Now, it is artificial intelligence. And educators must ask: What does this innovation help students learn that they couldn’t without it?

Unlike previous classroom technologies, generative AI can perform the very thinking that many lessons are designed to develop. It can write essays, solve problems, summarize readings and produce analyses that students are meant to do themselves. 

 have now issued AI guidance for schools. Much of it centers on data privacy, human oversight and teacher training, rather than whether a given use of AI is developmentally appropriate for a child.

That leaves the question most states have been skipping: What thinking should students do themselves? To answer that, it helps to look backward. In 1958, psychologist B.F. Skinner designed a  around a simple principle: Students could not move to the next problem until they had produced the correct response themselves. The machine provided immediate feedback, but it never completed the work for the student.

Whatever one thinks of Skinner’s broader work, he understood that students learn by responding to problems, answering questions, making mistakes, receiving feedback and trying again. The student must do the thinking.

But today, generative AI can complete the assignment without the student.

This is referred to as cognitive offloading, handing one’s thinking off to an external system. A  involving about 1,000 high school math students tested this, comparing two versions of an AI chatbot. One answered questions outright. The other offered hints and guidance, but not the answer. Students with the first chatbot performed worse on later exams than those who did not use AI at all. Students who received only hints and guidance performed as well as the no-AI group. One tool did the thinking for students; the other made them think first.

The rule that should anchor any school AI framework is simple. Its role should expand only as a child’s own foundation grows.

In kindergarten through second grade, students are building the skills everything else depends on, including sounding out words, forming letters and developing number sense. How well a child masters these literacy skills  in third and fourth grade, and  is indicative of math performance years later.

Children build those skills by doing the work; such as sounding out words, making errors and trying again. If AI gives students the answer, that productive struggle never happens, and the skill never truly forms.Ěý shows that the more students hand a task to a machine, the less able they become to do it themselves.Ěý

So, in K-2, AI belongs only in the teacher’s hands. There, it can help educators in planning phonics instruction and generating practice at each student’s level.

Third grade marks the shift from learning to read to reading to learn. The same shift happens in math and writing, where the foundational skills become tools for building arguments, drawing inferences and working through problems. That’s what makes a different kind of AI use possible. 

In grades three through five, students should use AI only after they’ve done an assignment themselves, and only under a teacher’s direction. A teacher might take a paragraph a student has written, generate an AI version, and have the class pick out errors and find weak reasoning, wrong facts and claims that do not hold up.

Working with flawed examples is one of the oldest ways to teach analysis, and AI produces an endless supply of them. A  found that when students critique AI-generated work with a teacher’s guidance, their critical thinking improves. The same research points the other way, too; the more readily people trust AI’s output, the less they think for themselves.

By middle school, AI should become something students should study, not just something they use. This is the age when the judgment to evaluate information .ĚýEvery eighth grader should be able to explain how these technologies work, where they fail and why they can produce convincing but wrong text — the beginning of true . Students’ own work should still come first, with AI used afterward and under a teacher’s direction.

By high school, students who can read critically, write clearly and work through problems independently can get real value from AI. In writing, they can first draft an essay, then use AI to help identify gaps or missing arguments. The students would then weigh the feedback to decide whether to follow the tool’s suggestions. AI could also work as a tutor, helping students, say, catch up in algebra or progress beyond grade level, giving feedback faster than a teacher with 25 students can.

High schoolers are also entering a world where AI is embedded in most professional fields, and they need to learn to use it well. That means having the capacity to think, write and reason without it, direct the tool and catch its errors. Those skills are the prerequisite to AI readiness. 

One principle should hold across all grades: Keep assessments AI-free. Tests should measure what the student can do, not what a chatbot knows.

Technology has never been the goal of education. Learning is. Schools exist to help students develop knowledge and skills they couldn’t develop on their own, and every technology should earn its place by helping accomplish that. AI may become one of the most powerful instructional tools schools have ever adopted. But its success shouldn’t be measured by what it can do. It should be measured by what students can do as a result of it.

Anthropic, one of the world’s most prominent artificial intelligence companies, is launching a version of its AI-powered assistant Claude for teachers, entering a race by technology companies to infuse AI into education.

Anthropic boasts that its product can incorporate academic standards from all 50 states, and teachers can use it to help devise lesson plans, personalize instructional materials to students, and harness data to improve instruction, according to a company news release.

Claude for Teachers joins Google, OpenAI, and Khan Academy — among others — in marketing AI products specifically to K-12 educators. The new product launched Tuesday and is available for free for verified educators in the U.S. It will also be piloted in Detroit Public Schools Community District for a study on educator well-being and practice.

Claude’s formal arrival into the classroom comes during a complicated moment at the intersection of technology and education. The , the , and have all encouraged educators to adopt AI. But a is simultaneously gaining momentum, prompting some of the nation’s largest school districts to rethink how much time students spend in front of screens, as well as the contracts they’ve signed with huge players in ed tech.

Drew Bent, education lead for Anthropic, said that teachers using Claude’s educator product could, for example, pull in a student’s past assessment data and assignment data, along with past lesson plans, and ask Claude to build lesson plans for individual students based on that data — all while they’re sleeping.

In developing Claude for Teachers, Bent said Anthropic staff often heard that while teachers are already using AI to generate lesson plans, the plans generated were often detached from the content teachers actually needed to address. Anthropic’s tool will help teachers save time and toil less to improve student outcomes, he said.

“There’s a lot of evidence of what works well for teachers in terms of aligning with high-quality instructional materials, formative assessments, differentiated instruction,” he said. “But of course, if you have 30 students in your class, you’re not able to do all of that.”

Bent said Detroit was already using other Claude products, and in a “human-centric” way that impressed Anthropic, leading to the pilot program in the district that will start next school year. Anthropic will train teachers at a handful of schools in Claude for Teachers, and evaluate how the product may shape teaching practices in the district.

While AI companies have been eager to cement the technology’s status as a classroom staple, tech giants likely have a long way to go to quiet skeptics. Student-facing AI and so-called cognitive offloading, a reference to the reliance on AI to complete tasks instead of using critical thinking skills.

Bent emphasized that Anthropic is focusing largely on teachers, and that most K-12 students can’t access the company’s Claude assistant, due to an age restriction for anyone under 18.

Daniel Buck, a research fellow at the right-leaning think tank American Enterprise Institute, argued that if teachers outsource work to AI, “don’t be surprised when classroom community and academic outcomes rapidly deteriorate.”

Skeptics have also raised questions about student privacy in using AI, such as Claude.

Anthropic is working with the American Federation for Teachers on aligning the product’s privacy practices with what the labor union has said will become a “gold standard” in best practices around safety, according to the news release.

Among the privacy features in Claude for Teachers: It won’t use conversations between the AI assistant and teacher accounts to train its AI, student information will be protected in a manner built to comply with the federal law governing student privacy, and privacy terms of service are written without jargon so teachers can understand what they’re signing up to use.

“It’s important that Anthropic is committing to these principles in their new Claude for Teachers — a tool designed by and for educators to assist them instructionally and hopefully give them more time for the human relationships at the heart of learning,” wrote AFT President Randi Weingarten in the company’s press release.

Weingarten has been walking a tightrope when it comes to AI. She’s in elementary grades while promoting teacher training in the technology. Just a day before calling for the ban, she , an AI chatbot from Khan Academy, in action.

While AFT is working with OpenAI, Microsoft, and Anthropic on privacy standards and AI training for educators, it is notably not working with Google, .

Utah’s state education board recently made a deal with the tech giant , promising personalized instruction tools for educators.

It’s not yet clear whether one AI product reigns supreme in schools, but more teachers overall are using AI. Around 61% of teachers , compared with 32% in 2024.

Chalkbeat is a nonprofit news site covering educational change in public schools. This story was originally published by Chalkbeat. Sign up for their newsletters at .Ěý

“I’m not going to pretend like you aren’t,” he tells his students. “I know it’s readily available for most of you, if not all of you.”

A math teacher at in Fullerton, California, where he works with students as old as 19 who are struggling to get enough credits to graduate, Rabanera has watched AI creep into homework assignments over the past few years as students use powerful tools like ChatGPT and Google’s Gemini to race through assignments. 

It has forced him to change his approach. 

He has stopped sending home problems that can be lifted wholesale into an AI chatbot and pasted back into an assignment. Instead, he builds lessons around what students actually care about, creating, for instance, a unit on buying a car that weaves together calculating interest rates and monthly payments with learning about credit scores. He has replaced rote problem sets with one-of-a-kind poster projects and in-class design challenges. 

When Rabanera assigns practice, he often has students devise their own word problems around personal interests to prove they understand the underlying concepts.

He’s hardly the only one scrambling to try something new: Nationwide, teachers at all levels are rethinking, scaling back or, in some cases, abandoning homework altogether as evidence mounts that students who outsource their assignments to AI aren’t just submitting work that isn’t theirs. They’re surrendering the cognitive struggle that makes learning stick and makes homework, well, work.

New large-scale research suggests that teachers’ fears are valid. A led by Sina Rismanchian of the University of California, Irvine, analyzed 3.2 million student math problems on the digital platform over a decade and found that after ChatGPT’s release in late 2022, high school students spent 31% less time on word problems — the kind easily copy-pasted into an AI — compared with graph-based problems that required a hands-on interaction with the platform. College students showed a 27% decline. 

When students were tested under proctored conditions with no access to AI, the copy-paste behavior vanished. And when researchers examined whether students had actually retained anything, they found that the odds of correctly answering AI-susceptible word problems fell by 25% in the post-ChatGPT years.

“Students are using AI a lot,” Rismanchian said in an interview. For those who do, “it’s coming at a cost for their learning outcomes.”&˛Ô˛ú˛őąč;

The College Board last fall that the percentage of high school students who said they use AI tools for schoolwork grew from 79% in January 2025 to 84% in May 2025.

In a , 37% of K-12 principals said students were using AI for homework help, slightly higher than the percentage who said students were using it to help draft essays.

John Singleton, an associate professor of economics at the University of Rochester and a co-author of the study, said the finding “certainly requires a rethinking of what the object of homework is.” For him, assigning short writing assignments to his college students is “insane these days, because you’re going to get back 25 AI-generated short essays, and so it’s really not gauging comprehension. It’s not even doing the work of forcing the student to engage with the material, because they can put it into the AI.”

Conscientious instructors are drafting AI policies that they post to class syllabi, he said, “but I think the temptation is there.”

Talking to colleagues, Singleton said, “Everyone feels sort of bewildered about whether they’re doing the right thing.” Moving toward presentations and oral exams make sense, but giving up traditional writing assignments as a way to assess student thinking, he said, “is too bad, in some ways.”&˛Ô˛ú˛őąč;

The irony of this moment is that AI was supposed to offer students a , capable of explaining concepts, adapting to individual learners and helping them work through difficult material at their own pace. Instead, many educators say, for a significant share of students it has become the most efficient cheating device ever invented.

“There’s a zillion people that are trying to come up with these guided learning environments and Socratic tutors and stuff like that,” said Justin Reich, director of MIT’s Teaching Systems Lab and host of the AI-focused podcast “.” “And I’m just like, ‘Guys, you’re putting the “Carefully teach me this stuff” button directly next to the “Do everything for me” button.’ ”

Reich has spent years studying why students cheat. When they’re being honest, they typically tell researchers that the assignment didn’t seem worth their time, or that they ran out of time. They felt pressure to perform or, in many cases, they found themselves stuck on a problem with no other help in sight. 

“What’s new now is that, with all the gen AI stuff, the cost of taking a shortcut is zero,” said Eric Cosyn, a researcher who co-founded the .

Ashley Kannan, who has taught eighth-grade U.S. history for 30 years in Oak Park, Illinois, said that if schools continue to go down the same path of assigning work and expecting students not to be tempted to take shortcuts, “the war is over — we’ve lost.” Classrooms, he said, will be left in “a race to see who can plagiarize and cheat the best, and who has the resources to do so,” a dynamic he calls a losing bet for everyone. 

Start your homework in class 

In interviews, many educators and researchers were quick to point out that AI didn’t invent academic dishonesty. 

Denise Pope, a senior lecturer at Stanford’s Graduate School of Education and co-founder of , a research and school reform project, has been tracking student cheating behavior for two decades. Long before ChatGPT, she said, copying a classmate’s homework was consistently the most commonly admitted form of academic dishonesty.

The group’s latest academic integrity study, drawn from nearly 30,000 high school students, shows that this pattern still holds: 32.7% of students reported copying someone else’s homework at least once in the past month, a figure almost identical to the share who reported using AI as an unauthorized aid: 32.8%.

Students, Pope said, are simply swapping out one shortcut for another.

“This sort of hand-wringing that AI is changing homework like never before is a little bit off,” she said, “because there were high amounts of copying and cheating homework long before.”

All the same, Pope’s team has surveyed more than 100,000 students since November 2022, and the results are unambiguous: They’re using AI to do homework. Many don’t frame it as cheating, making the case that consulting an AI is no different than asking a parent, calling a tutor or typing a question into Google. Echoing Reich’s findings, she noted, “Some of them are saying it’s another piece of technology that helps us when we’re stuck.”

Teachers, naturally, see it a bit differently. A Challenge Success survey of 678 faculty and staff members found that the most pervasive concern, raised by about 58% of respondents, wasn’t cheating itself but the erosion of critical thinking. Teachers complained that students aren’t developing the intellectual stamina that hard problems require. “If school is about skills and not content,” one teacher wrote, “ChatGPT takes away critical thinking skills at a time that we are supposed to be teaching those skills to the students.”

Pope said her group is hearing from teachers that they’re afraid to send homework home “because it’s even more clear that there’s this ‘Easy’ button, and students are not having the productive struggle that they need to really learn the material.”

She recommends rethinking homework, starting with what she calls a homework audit — a systematic review of assignments to ask whether they actually require a student to do the intellectual work required. Teachers should also be able to tell if the work was done by the student or by AI.

“Start your homework in class,” she advised. “You will get a really good picture of who understands what you’re asking and who doesn’t by looking around and seeing what happens in the first 10 minutes — one kid is done and one kid is still stuck.”

Kannan, the Illinois history teacher, has landed on a similar idea, built around conversations. He still assigns a version of the same paragraph students have long written to identify a historical figure and place them in context — but the process now unfolds through one-on-one conferences rather than solo writing. Because of the conferences, the writing looks different. 

The goal, he said, is to locate the assignment “in the hearts and minds of a student” rather than in a generic prompt that AI can complete on command. “I think that there’s a way to personalize rigor,” he said. “When students, when young people feel that something is personal to them, they do come alive.”&˛Ô˛ú˛őąč;

Kannan said this kind of individualization isn’t new — special education teachers and speech-language pathologists have practiced it for decades. “All I’m suggesting is that there are pathways that we know work. Why not bring it into the mainstream classroom for every student?”

‘How this could this stronger?’

Researchers are also grappling with the limits of what they can measure. Self-reported cheating data, as Rismanchian’s paper notes, is “the least reliable way to measure anything.” In his own earlier research on 70 undergraduates, more than half of students who were directly observed relying on AI denied using it. “We were actually observing this copy-pasting behavior,” he recalled.

A group of McGraw-Hill researchers co-authored the Rismanchian study, and Dylan Arena, the publisher’s chief of data science, said solving the AI cheating problem has two prongs: Better detection helps, but it’s insufficient without a cultural shift inside classrooms. Students who believe their sole obligation is to produce a completed assignment, he said, will always find the path of least resistance. 

“There are kids who are here thinking, ‘I need to punch my ticket, I need to get this thing done,’ ” Arena said. “What’s the most expedient way to do that? Hand it over to this tool.”

But he suggested that something deeper is actually happening as AI colonizes students’ thinking: They’re losing their tolerance for “not knowing” something at any given moment, for what he calls the productive discomfort of sitting with a hard problem before the answer becomes clear. “It used to be that we would spend weeks not fully understanding something, and we would read, and we would write, and we would talk, and we would try, and we would write drafts — and they wouldn’t be quite right. But there is now an expectation that I either instantly know what I should do, or I should turn to a tool.”

A few teachers are experimenting with a more informal version of that transparency, built on relationships rather than documentation. 

Kannan described pulling aside a student last year whom he suspected, based on months of conversation, of leaning on AI for most of his schoolwork. “I knew that because I was taking the time to talk to him,” Kannan said. 

Rather than report the student, he asked him to run his own writing through the same chatbot he’d been using, in a bid to “reverse-engineer” and improve it. “Let’s actually ask questions to ChatGPT about how this could be stronger? How is this weak?”

Kannan has since built that move into his regular teaching: After students draft work with AI’s help, he sometimes has them ask the tool to critique the output, with students in effect “co-designing” assignments rather than simply handing them to a chatbot and turning in whatever pops out.

In the absence of such guidance, he said, students will do what they must to complete assignments. He recalled a student who’d been assigned an essay on the American dream by an English teacher. “There was really no instruction as to what that was, and she said, ‘I carried on a conversation with Gemini about what the American dream was, and that helped me understand it more.’ Given the demands of what the student was facing, she used AI as a partner, as an opportunity.”

, a longtime education researcher who has studied homework, noted that teachers have been assigning less homework for at least a decade — and that the rise of AI might reduce it further if teachers lose confidence that take‑home work is genuinely done by students. But it could also work the other way: If teachers believe that AI is a kind of all-purpose helper, they could actually assign more homework because they believe students “are going to be helped out and kind of semi-tutored,” he said.

Many school districts are trying to reframe the relationship between students and AI from the ground up. The Laguna Beach Unified School District in California, working with researchers at Stanford, found that AI policies had generated a culture of suspicion in which teachers spent their energy trying to catch cheaters — and students felt guilty for using AI.

Michael Morrison, the district’s former chief technology officer, called it “the absolute worst culture that I can think of.”&˛Ô˛ú˛őąč;

In response, the district developed an add-on tool for Google Docs called , which asks students to disclose exactly how much and in what ways they use AI on a given assignment — a kind of nutritional label for AI-assisted work. Surveys suggest students generally use it honestly because the tool gives them something they didn’t have before: a sanctioned way to tell the truth.

Michael Keller, Laguna’s director of social-emotional support, noted that the district is already rethinking homework altogether for high school students, since 70% are athletes who devote an hour or more each day to practice and training. 

He sees the AI initiative as part of a broader commitment to treating students as full partners in their own learning. Monitoring how students are using AI, he said, is not the point. “We really view it as our moral obligation to make sure that we put trust and supportive relationships as the foundation to their learning experience,” he said.

For MIT’s Reich, transparency is necessary but not sufficient. The deeper challenge is motivational. “Kids are just natural boundary pushers,” he said.

And though they may not be able to articulate this, “the boundaries are what make them feel safe and loved and cared for.”

The problem, Reich argues, isn’t that students are weak-willed or morally deficient. It’s that the incentive structure of homework — grades for completion, not for thinking — has always been fragile. AI has simply exposed that fragility. 

“It’s not like homework is perfect,” he said. “Any teacher will tell you that some of the assignments are dumb or don’t work. But across the tens of millions of minutes of stuff that we ask kids to do in the afternoons, some of it’s got to be useful. And if we turn that spigot off, there’s just going to be less learning.”

That’s why our newsroom recently launched a new on Instagram () spotlighting a handful of important articles, case studies, data analyses and solutions snapshots that might have gotten lost in the noise. (You can also get these kind of daily highlights delivered straight to your inbox by signing up for our national newsletter.)Ěý

From the new demands facing college students in this era of artificial intelligence to a leadership transition at America’s largest union, here are five stories you might have missed this week: 

1. Elementary Principals Are Getting a Crash Course in How Young Kids Learn Best

Young children need different conditions than older students to thrive. Some school leaders are adapting their practices to reflect that reality.

2. Princess Moss Elected NEA President With Votes From 50% of Assembly Members

The former music teacher, who previously served as the union’s vice president, narrowly avoided a run-off in July 5 election.

3. What Today’s College Students Need That Previous Generations Didn’t

College degrees don’t guarantee good jobs anymore, so students must build resumes and AI skills in high school and college, experts say.

4. A Restored Detroit Rail Station Looks to the Future, Offering Training for Teens

Michigan Central Station, long a symbol of decline and decay, is now an innovation hub with workshops and cutting edge equipment for youth.

5. Q&A — ‘Rehumaning’ Education: Banning Screens Is Only Part of the Solution

‘We don’t have to remove all tech,’ author Stephanie Malia Krauss says. ‘We need to remove toxic tech’ and bring back play and adequate sleep.

But recently I have noticed something I never expected.

Many of my more thoughtful and honest students are becoming critics of the very technologies that shaped them. They are tired of the slop. Tired of content created by code. They say they are repelled by the prospect of an AI friend or romantic partner.

When it comes to the classroom, they willingly admit they prefer a lively class discussion to an online activity. They resent teachers who use AI to grade their papers. And most powerfully of all: They admit their use of technology is a hurdle to becoming the educated Americans they know they should become.

They speak almost like addicts.

They don’t want to be on their phones eight or nine hours a day. They don’t want to use AI to complete their assignments and short-circuit their ability to learn and grow. They know their attention span is stunted.

But in so many circumstances, they simply can’t resist. These observations may sound anecdotal. Let me assure you, they are not.

Over the past year, a series of highly publicized incidents have suggested the emergence of something larger: the first widespread cultural backlash by young Americans against the digital world that shaped them.

Multiple public speakers recently referenced the AI revolution now upon us—a at the University of Central Florida, a at Middle Tennessee State University, former at the University of Arizona. In every instance, young Americans either passionately booed or, in Schmidt’s case, the remarks.

A community college in Arizona to read the names of graduates, but the system quickly malfunctioned. When the college announced what had happened, the backlash was both raw and immediate.

These incidents may appear isolated or trivial.

They are not.

Together they suggest the emergence of something larger: the first widespread cultural backlash by young Americans against the digital world that shaped them. The generation that grew up on iPhones, spending much of every waking hour online, now seems to be awakening to the perils of a digital world neither they nor their parents fully understood.

And yet these young people are increasingly lending their voices to a growing chorus of educators, parents and policymakers who have begun to realize a painful truth: There is no quick technological fix for the crises consuming the modern American classroom.

, , emotional distress and a generation-long erosion in students’ ability to concentrate all demand a fundamental reassessment of the role screens now play in the educational lives of our children.

Every other day a prominent newspaper or publication now gives voice to this fundamental truth.

Whether it’s The New York Times explaining or prominent Substack columnists offering a American educators should take advantage of this moment by defending traditional instruction rooted in foundational human relationships.

A generation ago, young Americans had access to a diverse chorus of influential adult voices that tethered them to the mighty responsibilities and possibilities of adult life. Children lived with two parents, numerous siblings and often spent considerable time with grandparents. Life exposed young minds to a faith tradition with pastors and priests on Sundays, sports activities with coaches after school and maybe Boy Scout or Girl Scout leaders as well.

Many of these voices from ages past are silent today.

But that doesn’t mean our children don’t hear voices. They do, and they often come from people and digital spaces their parents never would have chosen for them.

This is why — especially in this era — the humanity of teachers and the personal vitality of our classrooms are essential. The voices of instruction our children hear should be the voices of teachers who know and care about their students — not Alexa, not Siri, not some anonymous digital interface designed to maximize engagement rather than human flourishing.

Eye contact. Conversation. Personal relationships. No code or product required.

The kids know this. The kids want this.

Shame on us if we fail to give it to them.

Your host in Osaka, Japan, slips on a and suddenly hears your words transformed into flawless Kansai Japanese. Even better, their reply in their native tongue comes through perfectly clear to you.

Has your kid ever used artificial intelligence to answer homework questions that their teacher used artificial intelligence to write?

It’s possible, according to a  from the Georgia Department of Audits and Accounts that found a majority of teachers in the state are using generative AI for planning or in the classroom, but also express concern that students’ use of the technology could harm their learning.

Teachers see promise in artificial intelligence

The poll, based on more than 13,000 teacher responses from across the state, found that 59% of those who responded said they use AI for teaching tasks.

Teachers were most likely to say they used AI to prepare for class — 95% of teachers surveyed said they used it for instructional planning and preparation at least a few times a year, with more than half using it at least once a week.

Venecia Whyte-Foster, a middle school English teacher in Savannah-Chatham County, said she was an early adopter of the technology.

“I embraced it immediately as it came out,” she said. “I’m like, ‘OK, this is not going away. My kids are going to use it. How do I use it in my class effectively?’ ”

Part of what she came up with involves personalized chatbots that turn classroom concepts into escape room-like games where students use what they’ve learned to solve puzzles and progress through levels.

“They have to ask a question directly toward the topic, and then the chatbot will give them more questions, give them information and then ask them questions, and until they’re able to answer those questions effectively, they cannot move on,” she said. “And they love those escape rooms, they love to be the first to get out of the room.”

The task the polled teachers were most hesitant to use AI for was grading student work, with 62% saying they never do that.

Of the teachers who use AI, nearly 90% said it had a positive impact on their classrooms, with majorities saying the technology helped save time, improve their lesson materials and increase student participation.

Teachers who do not use AI told state analysts they believe it invites students to become reliant on the technology, potentially damaging their critical thinking skills and encouraging them to cheat or become lazy. Teachers also expressed concerns about AI accuracy, data privacy and other ethical concerns.

University of Georgia associate professor Xiaoming Zhai directs the university’s AI4STEM Education Center, which focuses on using AI to advance STEM education.

Zhai said he understands teachers who don’t trust AI, but he still encourages them to keep an open mind.

“Regardless of whether they’re using AI in their own profession, their students in the future will be using AI in their career,” he said. “So if they’re not able to use AI, how can they expect them to educate the future workforce to be able to use AI responsibly and ethically?”

Zhai said teaching future educators how to effectively use AI will help them guide students toward the proper use.

“I think the key part is how can we guide a teacher to be responsible and ethically using AI so they can foresee any potential risk and challenge that the students might be facing when using the AI?” he said.

Middle and high schoolers are also embracing AI, to teachers’ chagrin

Elementary school teachers told auditors that their students were the least likely to use AI to complete their work, but adoption of AI for school use grows as students age. A majority of high school teachers report their students using AI on assignments at least half of the time.

Teachers whose students used AI were more likely to say that the technology had a negative impact on their learning, with a majority of middle and high school teachers reporting that AI had a very negative or somewhat negative impact on their students’ learning.

More than half of the teachers surveyed said they required in-class work or writing as a means of managing student AI use. Half of the respondents said they made an effort to clearly define AI abuse and what constitutes cheating, and 43% said they use software or apps to check for AI use in student assignments.

Whyte-Foster said she teaches about AI hallucinations — the word for when AI spits out a confident-sounding but totally fabricated response — and why it’s important to double check everything.

“We discuss why it’s important that we understand that it is not something that we just go through, but that we have to use our minds, that we have to put things into place to make sure we are thinking for ourselves, that it’s giving us information so that we can use it in our thinking, as a part of our brainstorming,” she said.

Despite that, Whyte-Foster said students sometimes try to pass along AI-generated work as their own. She said it’s usually easy to spot since she is familiar with her students’ writing abilities from their in-class work. When she gets an AI-produced assignment, she has a talk with the student about academic integrity and offers the chance to re-do the work for a reduced grade.

“They’ll tell you, yes, they’d prefer a 70 to a zero,” she said. “But if they get the 100 or the 90, I’ll put it there, and then put the minus 30 there and just say, ‘Hey, look at this, you did it, you could have done it before. The only reason you have a 70 is because you didn’t follow the guideline in the first place. Next time, choose the right way.’ ”

Guidance and training

About two-thirds of teachers reported that they had received guidance on AI use, while 70% reported that they had received training and 27% reported that they received neither training nor guidance.

Auditors noted that responses were inconsistent within districts, with a portion of teachers reporting they had received training or guidance and others reporting they had not. Auditors suggest that could indicate teachers were unaware of district resources or that guidance was provided to some teachers but not others. Teachers in districts with high poverty levels were more likely to say they had not received guidance or training.

The use of AI in K-12 education is not yet addressed in federal or state law, but the Georgia Department of Education and the Georgia Professional Standards Commission, the agency responsible for licensing teachers, have developed guidance.

In January 2025, the state Department of Education appointed an AI ethics and impact officer and released  best practices. In a statement, State School Superintendent Richard Woods said AI use should come with “serious consideration of the ethical, instructional, and privacy implications.”

“What I never want to see is AI being used to replace the skill and expertise of our teachers, or our students’ ability to develop critical thinking skills,” said Woods, a Republican who is running for re-election in November. “AI should always be a tool, never a replacement.”

That experience stayed with me long after graduation. It taught me something that is becoming increasingly vital in today’s digital world: Confidence is not developed by having the perfect answer but by believing in authentic ideas. 

Many young students are losing confidence in their own ideas. It is undeniable my generation grew up behind screens, with videos and phones embedded in everyday life. As a 16-year-old high school student, I don’t consider artificial intelligence a futuristic invention but rather a simple extension of the world we already live in. That is why I believe many conversations about AI in schools are missing the bigger issue entirely. The real concern is not technology itself but what happens when student voices become overshadowed. 

Technology has not only changed the way students learn but also the way we communicate on a daily basis. Many young people have become highly skilled at digital tools, with FaceTime and Zoom becoming frequent parts of everyday life. It is a common joke that if our generation does not put their phones down, they will forget how to talk to someone on a date! There is immense truth behind that. As we become more accustomed to immediate responses, we are slowly losing the patience to sit with a thought.Ěý

About 54% of students already use AI for schoolwork, a number continuing to rise. AI can certainly help students organize their thoughts. I even used it during the brainstorming process of this piece, but the important aspect is that my writing remained a reflection of my voice. With this expanding access to knowledge, an important question remains: Where does student voice fit in? If students begin relying on AI from the moment their education begins, they risk losing the discomfort that comes from developing confidence in their own ideas. That uncomfortable part matters. 

I have personally experienced how being in an uncomfortable position can lead the mind to function in ways AI cannot replace. It was the day of a long-awaited DECA business conference, and I put on my dress and blazer, a stark contrast to the comfortable clothes I wear at school. I had recited my speech until I knew it by heart, yet during the presentation, I felt the pressure set in and my thoughts begin to blur.Ěý

For a moment it felt as though all my preparation disappeared, but after taking a moment to slow down, I looked back at my notes and continued. It was not the polished performance I had imagined, but by the end I had conveyed what I wanted to say. More importantly, the experience taught me something AI never could, which was how to recover in real time.

Through these experiences, I have seen how much students can grow when they are asked to use their own voices. Recognizing this, I am proud to lead a summer program in the Chicagoland area called First Voice Academy. 

Here, middle school students learn and practice public speaking in a low-stress, immersive environment. Designed to promote interpersonal connection, the program walks participants through the importance of communication and concludes with their delivering a speech on their own. Taught directly by high schoolers, the program gives younger students the opportunity to learn from peers who have faced many of the same obstacles.

I can envision a student struggling to present during a speech, but that is the key: The words are theirs. The goal is not to ignore AI, but to ensure younger generations have opportunities to develop confidence in their own ideas.Ěý

Artificial intelligence can polish language perfectly and respond to a question in under a second, but what it cannot do is replace a student’s voice. Confidence is gained when students believe their own thoughts are worth sharing, even when they come out wrong the first time. Students who feel they have a voice at school are seven times more motivated to learn. 

A program like First Voice Academy allows for real world experience while in a learning environment. If similar programs can be expanded into schools, more students would build interpersonal skills at a crucial point in their lives.Ěý

I think back to the moment I stepped up to give my graduation speech and how scared I was. All it took was to say the first word, and suddenly I felt connected with the audience. Student voice offers a perspective unlike any other, and simply needs the opportunity to be heard. Technology may help develop ideas, but true opportunity comes from nurturing one’s own voice.

New York City education officials are hitting pause on releasing comprehensive artificial intelligence guidelines after their draft policy from March sparked .

Officials initially said their final guidance would be released in June, but backed away from that timeline during a Wednesday City Council hearing focused on AI in schools. Instead, the policy guidance will be released sometime this summer, First Deputy Chancellor Danielle Giunta said.

Giunta said the “shifting national conversation, which has really escalated over just the last couple of weeks alone” was one reason for the delay, along with nearly 6,500 comments responding to the March draft. Officials have not released those responses but told City Council members they would do so.

Worries about AI — and education technology in schools more broadly — have increased in recent months. More than half of City Council members urging Mayor Zohran Mamdani and schools Chancellor Kamar Samuels to pause AI use in schools citing concerns about learning and mental health. A broader petition seeking a moratorium on generative AI use in city schools has received .

The Education Department’s draft policy used a that spelled out various uses of AI and their risk levels. Assessments and grading are not allowed, for instance, but brainstorming lesson plans gets a green light. Student AI use was largely left out, even though it’s one of the most pressing issues schools face. In response to the draft policy, parents and largely criticized the approach.

Samuels, who did not attend Wednesday’s hearing, that the Education Department’s draft policy “missed the mark” that AI “is the most invasive technology that we’ve seen.” He suggested the final guidance would have stricter limits for the city’s youngest students.

Giunta said Wednesday that the city is considering different expectations based on age and grade level and thinking about how to prepare older students “for a world in which AI is already present without allowing AI to replace their own thinking.”

City Council members repeatedly pressed Education Department officials about the extent to which AI tools are already used in classrooms and indicated families should have a greater say over how their children are exposed to it.

“There are huge gaps in our understanding of how the technology is being deployed and when,” said Manhattan council member Carmen De La Rosa, who chairs the technology committee. “We do need the time to be able to wrap our arms around what is happening in our classrooms.”

Education officials said they are working with superintendents and principals to develop a more comprehensive sense of what tools schools are already using.

The Education Department’s stance on AI has whipsawed in recent years. When ChatGPT gained widespread use, city officials . Three months later, they rescinded the ban.

Former schools Chancellor David Banks , saying “AI can revolutionize how we function as a school system” and could be used for everything from college advising to assessing student work. But he never released a clear policy, leaving that task to Mamdani and Samuels to complete. Meanwhile, schools have been left to .

Chalkbeat is a nonprofit news site covering educational change in public schools. This story was originally published by Chalkbeat. Sign up for their newsletters at .

after laments the death of entry-level hiring, and according to the World Economic Forum, the they’re building will be out of date by the end of the decade. 

As both a mother of two teenagers and the head of one of the nation’s largest education , I’m asked a version of the very same question from both parents and policymakers: How can we ensure that today’s students are learning things that are actually going to matter?

I found part of the answer in an unexpected place: a high school theater.

My daughter recently performed in her school’s production of “Legally Blonde: The Musical.” She played Pilar, a scene-stealing member of Elle Woods’ sorority. After the show, she cried — not from exhaustion or stress but from pride. And not pride in the accomplishment, but the learning that came with it.

Of course, what she learned from that play won’t show up on a standardized test. Colleges won’t find it on her transcript, either. But like thousands of other theater kids, she’s building exactly the sort of skills that the labor market is demanding.

In the weeks leading up to the performance, she had to collaborate with a diverse cast, manage her time across competing priorities, take direction and feedback, recover from mistakes in real time and perform under pressure. She built confidence, resilience and the ability to communicate with clarity and presence. We used to call these “soft” skills. In the age of AI, they’re the hard currency of economic mobility. 

Perhaps it should come as no surprise that some kids who participated in the performing arts have landed themselves in positions of significant power and influence, including former Disney CEO and U.S. Supreme Court Justice .

As AI accelerates the automation of routine tasks, a suggests that the uniquely human capabilities are becoming even more valuable, not less. More often than not, these “durable skills” are cultivated in places we don’t traditionally count: theater productions, debate teams, student government, part-time jobs and community-based experiences. 

This raises an urgent question: If the skills that matter most for success in the world of work — and the world at large — are developed in unconventional ways, why do we continue to treat those experiences as peripheral?

The simple answer is that our education system is largely oriented around what is easiest to measure over what matters most. If we can’t measure it, we can’t evaluate it. And if we can’t evaluate it, why teach it? But when it comes to high school, what we measure is beginning to change.

In recent months, a growing number of states are changing high school graduation requirements and replacing traditional diplomas with “,” designed to provide a broader understanding of what students have learned that more holistically captures the skills necessary for success in life. 

The Carnegie Foundation recently a new set of skills progressions designed to complement “decode” the “skills genome” by transforming skills like collaboration and critical thinking into their component parts — a significant step toward developing new forms of assessment, curricula and ultimately teaching methods that bring “theater” skills to center stage.

As business leaders begin to question the value of longstanding skills-proxies — including even the college degree — they are signaling to young people that skills honed outside the academic context are not optional; they are essential. That includes designing hiring and interview processes that explicitly take human skills into account; assessing those skills as workers progress in their careers; and adopting training programs that focus on those skills alongside technical competencies.

 More and more businesses are recognizing that if the pace of technological advancement isn’t slowing down, the best way to keep up is to ensure that their employees have the resilience and agility to navigate a world of work defined by change.

And for parents, perhaps most importantly of all, it may mean recognizing that the path to opportunity is not always linear or confined to the classroom. It’s time to stop thinking of theater, sports and volunteering as ways of burnishing a resume for college. Those activities have always been the places where students learn to work together, navigate uncertainty and step up in scary situations. Reading and math will never stop being important, but without the skills to put academic accomplishments to work, too many of our young people will find those dire headlines starting to come true.

My daughter didn’t just perform in a musical. She practiced the very skills that will help her navigate a world where change is constant and careers are nonlinear. If we are serious about preparing young people for the future of work, we need to expand our definition of what counts as learning — and where it happens.

Appalled, Jessie turns around to find the twins’ discarded toys in the yard, looking worn and weary. 

“What happened to you?” Jessie asks them. 

“Tech,” the toys reply in unison. 

These toys explain to Jessie that, for years, technology has been “invading” homes, capturing kids’ attention and effectively ending imaginative play. 

“Your kid still plays?” one asks. “I don’t remember play.”

Jessie is skeptical, so they prove it to her: The toys climb to the roof of the twins’ house and look across the neighborhood, where they can see the blue light from screens glowing back onto children’s faces through each window.

“No wonder she can’t make a friend,” Jessie says, referring to Bonnie. “She’s the only one out there still playing with toys.”&˛Ô˛ú˛őąč;

It’s the type of exchange that stays with you â€” perhaps because, when depicted in a fictional animation, it’s clear just how representative it is of our current reality. 

The latest film in Disney-Pixar’s multibillion-dollar Toy Story franchise, which was released on June 19 and pits traditional toys against digital technology, comes at exactly the right time, multiple child development and media experts said in interviews. Children, even in their early years, are awash in screens and devices, and their overwhelmed parents often don’t know how to go about establishing guardrails — or what those guardrails should even be. All of this comes as AI is emerging as a threat to children, now embedded in toys and being used to generate videos that are served to children on YouTube without any real quality checks. Meanwhile, a chorus of critics — from the to the of one of the largest teachers unions in the country — are adding their concerns to the conversation.Ěý

“I think this is one of the most important issues of our time. I really do,” said Kathy Hirsh-Pasek, a professor of psychology at Temple University and senior fellow at Brookings Institution who studies child development. 

“We are at this inflection point,” she added, “where we actually have a technology that has the capability, if used incorrectly, not just to captivate kids’ attention but lessen kids’ attention. It takes away human-to-human interaction by displacing time and their imaginations.”

Indeed, “Toy Story 5” addresses exactly that. Soon after Jessie learns what’s been happening to children all over the neighborhood, 8-year-old Bonnie gets a tablet of her own, a frog-shaped device called Lilypad. 

Lilypad instantly grabs hold of Bonnie’s attention. Bonnie — a child who had previously relished in the rich stories and scenes she’d created with her analog toys using just her imagination — becomes zoned out, using the device day and night. She forgets about her beloved toys and the worlds she had dreamed up for them. 

Bonnie’s parents bought her the device in hopes that it would help her make friends. After all, everyone else her age seemed to be using one. While Lilypad does help Bonnie connect with her peers through a group chat, she soon learns that’s not quite the same thing as finding a true friend. After getting invited to a sleepover that goes all wrong, Bonnie finds out that digital connection is no replacement for authentic human connection. The child becomes more and more despondent and detached as the film progresses, leading her toys — Jessie, Woody, Buzz and the rest of the gang — to take charge on her behalf. They want to oust Lilypad and also help Bonnie make a real friend. 

The movie portrays technology as a force that can come at the expense of human relationships, which is often what experts warn about when it comes to children and screens. 

“We see human displacement as a very real threat,” said Michelle Culver, founder of the Rithm Project, a nonprofit working to reclaim and evolve human connection in the age of AI. So she loved that “Toy Story 5” was, in many ways, “a celebration of friendship.”&˛Ô˛ú˛őąč;

Dana Suskind, a pediatric surgeon and author of the forthcoming book “Human Raised: Nurturing Connection, Curiosity and Lifelong Learning in the Age of AI,” felt the movie did a good job of demonstrating “how screen time can erode creativity and connection and lead to bullying.”

“Even though the movie is about tech,” Suskind said, “it’s probably about something deeper that we’re all grappling with: the desire and the need for human connection, the important role the early years play in navigating human relationships and learning how to have relationships.”

Suskind saw “Toy Story 5” the day it came out. She went with two colleagues at the TMW Center for Early Learning and Public Health at the University of Chicago, including Liz Sablich, who brought her two young children along. 

Sablich’s 7-year-old son Jack had some astute questions and observations about the film later that night, after taking some time to process what he’d seen, his mom said. One of the first things he said, Sablich recalled, was that Bonnie seemed “hypnotized” by her device — that was the word he used. “It was like she worshipped it,” Jack told his mother. 

Sablich agreed and explained that some technology is actually designed to do that. Jack wanted to know why. When she told him that, in some cases, the more a person uses the technology, the more money a company makes, he became visibly upset and said, “That’s not right. That’s not fair. They should care more about the people.”&˛Ô˛ú˛őąč;

Jack was also perplexed that Bonnie kept her device in her bedroom, right next to her bed, at all times. (Jack is only permitted access to his tablet during travel, Sablich said.) He said he never wanted to use his tablet again and didn’t want his mom to use her phone. Sablich used that opportunity to explain how technology can work for and with us, too. For example, she was able to use her phone to arrange a play date for him the next day. “We can use it in limited ways to make it easier to play with friends and do things in the real world,” she told him. 

“I was pleasantly surprised by how straightforward I thought the message of the film was and that it sparked such an important conversation,” Sablich said. 

That’s exactly the type of dialogue experts are hoping the film will prompt. 

“We have to get this in the national conversation,” said Hirsh-Pasek. “The group that’s being left out of the national conversation is little kids, the 0 to 10 crowd.”

The film is giving families and young children an opportunity to “pause and reflect,” said Sunny Xun Liu, director of research at the Stanford Social Media Lab — and to ask some important questions.

“What are our values as a society for technology, especially for such young children?” Liu said. “What values should we hold together as a community and a culture about technology and screens? It gives us opportunities to think about this topic as families, communities and a society together.”

That Disney, a powerhouse media brand trusted by millions of American families, opted to tackle this topic at all was “bold,” Suskind said. Michael Levine, an early learning expert with more than two decades in the field of children’s media, called it “gutsy.”

“The fact they’ve responded to the cultural moment … is really smart and, I think, brave,” he said. “Hats off.”

Culver, at the Rithm Project, was impressed by how much “dimension” the movie had on the topic of technology, showing that, in addition to the harm devices can do, they can also help bring people together when used appropriately. 

“There’s always the opportunity for good,” Suskind noted. “It’s whether or not we make the design choices that actually allow it. I think Disney tried to show that.”

College grads are facing a tough job market, with headlines almost daily declaring their prospects or or call their a

High school students entering college need to think ahead and build work experience and skills as soon as they can, experts say. College is still key to long-term success. Just not on its own. 

It’s now what some experts refer to as “college and”: and an internship; and training in at least Artificial Intelligence basics; and a career credential; and some project or passion that demonstrates a student can use the skills learned in class in the real world.

“Durable” or “soft” skills such as teamwork, reasoning and collaboration are also considered a must in the workplace.Ěý 

 â€œWe’ve seen a transition into employers wanting a little bit more quantifiable evidence that students actually have skills,” said Scott Fleming, executive director of the State Council of Higher Education of Virginia. 

“Employers who are looking at first hires out of college… they want to see that they already have work experience,” Fleming told a . “The degree is important. But did you also get work integrating the learning, or an internship, project based learning, undergraduate research, something else as part of that education enterprise? That, to an employer, signals as well that you have developed those skills along the way.”

AI which is automating tasks, is also disrupting the job market, experts noted, on top of the increasing demand from employers for more evidence than just a college degree that graduates have the skills they want.

Jeff Strohl, director of the Georgetown University Center on Education and the Workforce, said he doesn’t envy students trying to chart a path toward a career with all the shifting demands of employers and the impact of AI on the economy.

“In the 1980s, you just got a BA, the employer would train you and everything would be nice,” Strohl said. “Now much more of that workforce preparation falls on the student.”

He still urges students to go to college, saying higher education degrees offers the best chance at long-term success. But he and several other career experts and researchers have a few key pieces of advice for young people as they prepare for a career:

• Learn how to use AI as a tool, not be replaced by it. 
• Build resumes while in high school and college with jobs and internships. Internships are scarce, but students still need to seek them out to build skills, contacts and references. 
• College majors can matter a lot, though AI is slashing the value of some once-certain degrees.
• Expect to be nimble and adapt as the job market changes.
• Consider how well a college provides AI training or work experiences when choosing a school.
• Find ways to show mastery of “soft” or “durable” skills through jobs, projects or new tests that are emerging that can certify mastery of them.

“Don’t get anxious,” Strohl said. “But be purposeful and deliberate… Engage, engage. Be deliberate. Don’t let stuff happen to you.” 

The data on today’s job market is bleak, with college graduates finding fewer jobs to launch careers, and many having to take jobs below what their degree would usually allow. The unemployment rate for recent college graduates was 5.7% in the first quarter of 2026, the Federal Reserve Bank of New York recently reported, the highest rate since the pandemic.Ěý

A separate study by the Federal Reserve Bank of Cleveland found the advantage a college degree has in finding a job over just a high school diploma — independent of the pandemic and AI. College graduates, researchers found, are now having to search for jobs longer than high school graduates do.

Between 1976 and 2000, high school graduates had to look for a job for 11 weeks before finding one, while college graduates found jobs faster, in just eight weeks. Today, high school graduates are still taking 11 weeks to find a job, but college graduates are now job hunting for 12 weeks.

“That… gap is now in favor of high school graduates,” said researcher Barış Kaymak.Ěý

It’s not clear how much the recent growth of Career Technical Education programs in high schools has affected the difference between high school and college graduates. Though students who earn career credentials in high school can have better job prospects after graduating and could draw some students away from college, the value of credentials varies greatly, and national data hasn’t shown enough students shifting to CTE to affect the trend.

The New York report also highlighted a larger problem: More than 40% of recent graduates were underemployed — doing jobs that don’t require a degree.Ěý

Underemployment is both costly to graduates and hard to overcome. A by the Strada Education Foundation and Burning Glass Institute found that underemployment gives graduates a “heavy financial cost.” Though college graduates earn 25% more than just having a high school diploma, even in jobs that don’t require a degree, that’s much less than the 88% more they would earn if they found a job that matched their degree.

“This leaves underemployed graduates on weaker financial footing as they start their careers, especially those with substantial student loan debt,” the report stated.

It’s also hard to recover if the first job out of college isn’t a good one. Strada reported that 73% of graduates who started out underemployed after college remained underemployed 10 years later.

Strada identified a few keys to improving odds of finding a good job, with the most important being picking the right major and having job experience before graduating. An internship or job, Strada found, reduces the risk of underemployment more than 49%.

Even STEM degrees, once thought of as a guarantee of success, aren’t created equal. Strada found that computer science, engineering or mathematics majors were much better off than those in fields like biology.

Degrees in nursing, engineering, architecture and math-intensive business fields such as accounting and finance had the best chances of avoiding underemployment.

A by the Center for an Urban Future offered a snapshot of what underemployment can look like by analyzing how graduates of the City University of New York system fared after leaving school:

That report found that:

• Job postings for applicants with little experience have fallen 37% since 2022
• Internship opportunities for college students are also down 37% since the pandemic.Ěý
• Ten percent of CUNY graduates were working in retail or food service five years after graduation, rising to 13% for graduates of CUNY’s community colleges.Ěý 
• Surveys show that just 12% of CUNY graduates had a paid internship during college.Ěý

“In this challenging hiring environment, CUNY’s mostly low-income students will struggle to gain access to well-paying jobs unless CUNY develops far stronger connections with employers,” the report stated. 

Merrill Pond, executive vice president of the Partnership for New York City, a leading business organization, also called out a need for more work-based learning.

“If students are not given the opportunity to put something on their resume that will make them stand out or start to develop that network, they’re often going to be at a disadvantage,” Pond said as the report was released.  

Though CUNY was praised for progress in connecting students to work, including through partnerships with several nonprofits that train and mentor students and help them find internships, the report found  efforts are still falling short.

Lauren Andersen, CUNY’s vice chancellor for career engagement and industry partnerships, said some of the report’s findings were “sobering.” 

She said the university’s own data shows 22% of students have paid internships, data it just started tracking in the 2023-24 school year. The university also started a new effort, CUNY Beyond, in 2025 to improve career advising and make faculty more aware of how they can help students connect to jobs.

Students are three times more likely to have a job at graduation if they do a paid internship, Andersen said, and twice as likely to have an internship or job if they meet with a career advisor.

Many students come to college without family connections to business or clear ideas of how to seek jobs.

“CUNY has to take on the role, and higher education institutions have to take on the role, of building those relationships for their students, and that’s something that we’re not necessarily historically set up to do,” she said.

Western Governor’s University is another school that has made a deliberate effort to connect students to jobs. It’s also joining a push by some high schools and nonprofits to evaluate and rate students on their soft skills to validate those skills to employers.

Students need to be able to show employers what they have learned and how they can succeed at work, Western Governors President Scott Pulsipher said at the same panel as Fleming. Colleges need to help them do that, as well as help them try out jobs to see if they like them and find out which skills they might need to do them. 

“Every rising high school graduate should be more intentional about developing their vocational identity — meaning, how do they see themselves contributing to the world?” Pulsipher said. “It feels to me like they defer this too long.”

The emergence of AI has created new complications for young people. Changes are happening so fast, it’s hard to plan what the job market might look like in a few years.

Georgetown’s Strohl said new high school graduates might be better off in four years than today’s college graduates.

“The shocks that we’re seeing to the labor market today, and are clearly impacting recent college grads… will be better understood,” he said. “The education system… will have had a chance to figure out some strategies of response, and actually, to what degree is AI going to be a problem. “

He also cautioned AI might not be as big a disruption as some fear. Automation had the “same kind of hype that we’re hearing about AI,” but worries 40 million jobs would be replaced by machines turned into eight million jobs at most just adapted to work with automation, not replaced by it.

Harvard’s McKittrick urged students entering college not to be caught up in the “noise” of a changing job market, with some firms hiring few people, but others hiring many. Along with working jobs while in college to build a resume, she said students need to learn to use AI.

“What we’re hearing from employers is that they do want young people with AI skills,” McKittrick said. “They might not exactly know what those skills are, but they want people who are kind of comfortable using the technology. Young people should be thinking about, ‘Does the college have an AI literacy strategy?’ ”  

Others said it will be key to also develop “soft” or “durable” skills such as critical thinking and ingenuity to put AI to best use.Ěý

“AI still needs the humans to say, ‘Even as I’m using it, how should I be using it?’ ” said Pulsipher of Western Governors University. “Are the outcomes that it’s providing me useful and relevant to what we’re trying to do?”

Laura Ullrich, Director of Economic Research in North America of the job search website Indeed, is telling her three sons to find jobs while in school to explore what they want to do, while also making sure to build soft skills that are becoming more important in job applications.

“I think a few years ago, you would have seen very few computer science majors feel the need to call out that they were a good writer or maybe had more critical thinking skills, or had taken a bunch of philosophy classes,” Ulrich said. “But today… young job seekers are thinking about this quite differently.”

Principals, superintendents and other school leaders are being subjected to sleek demos from presenters who know all the right buzzwords. Salespeople are showcasing technologies and hinting that, should administrators choose not to buy their product, students will be left behind. That the digital divide will grow — on their watch. 

This intense pressure will cause some school leaders to sign agreements that will not serve their students well. While most mistakes won’t rise to the level of prominent school districts supporting a failed startup, some administrators will nonetheless let anxiety rule the day. They will spend money on tried-and-true methods that could do more good than the shiny new thing. 

What can prevent this? Asking the right questions. 

I have spent years on both sides of the table. As a teacher, I used both good and bad ed tech, so I have seen what actually works in the classroom. As an employee of an ed tech startup I designed technologies that were sold to schools, so I know when promises are inflated to help the bottom line. As a professor of educational technologies and leader of the University of Southern California’s Center for Generative AI and Society, I study these technologies to know what works — and what doesn’t — when it comes to AI’s influence on education.

That is why, with the support from my colleagues, I have written a titled “Beyond the Salespitch: A Practical Guide to Questioning EdTech Vendors.” Below are some key questions every school leader must consider before signing on the dotted line and committing limited resources to AI-powered ed tech.

Does the Tool Really Help Teachers in the Classroom?

• What specific cognitive skills does the product aim to develop (e.g., argument construction, data interpretation, metacognitive reflection)?
  
• How does it supplement learning rather than replace it? Ask for a step-by-step walk-through showing where student input is required and how feedback loops are built in.
  
• What evidence exists that students improve on the targeted skills? Request pilot study results, raw performance data or an independent evaluation report — nothing less rigorous will do. 
  
  

Does One Size Really Fit All?

• Which student populations were included in testing the product? Look for data on age, language proficiency, special education status and socioeconomic background. Biased data means biased performance.
  
• How does the system adapt to individual students’ profiles? Ask for a live demo of the personalization engine that shows adjustments based on prior performance and teacher input. 
  
• What safeguards are there for students who might become over-reliant on the tool? Ask how the product detects and flags excessive automation, prompting teacher intervention.

Who Owns the Content?

• Where is student data stored and processed? Identify the physical location of servers (domestic versus overseas) and any third-party processors involved. Ask how data breaches are handled and communicated.
  
• Who owns the text or media students generate using the tool? Ideally, the contract must state that all user-created content remains the property of the school district. Otherwise, your school becomes another content generator for a company. 

• Will our data be used to train future versions of the AI? If yes, require an opt-out clause and a clear explanation of how opting out affects service quality.Ěý
  
• What happens if protected information (FERPA/HIPAA) is entered by mistake? Insist on a documented protocol for immediate deletion and breach notification. Consider clauses that allow you to exit agreements when large enough mistakes occur. 

Are Students Protected From Bias and Misinformation?

• What filters prevent misinformation or harmful content from reaching children? Get access to a live example of the system handling ambiguous queries that could lead to falsehoods. Stress-test it!
  
• Who is responsible when the AI produces inappropriate output? The contract should assign liability to the vendor for any harm caused by the tool’s autonomous decisions.

Can You Walk Away From the Contract?

• What are the cancellation terms? Look for a no-penalty opt-out clause that allows the district to discontinue use if outcomes aren’t met. 
  
• Is there an outcomes-based pricing model? Vendors that are willing to tie fees to measurable learning gains demonstrate confidence in their product.
  
• What support and training are included? Clarify the depth of professional development, response time for technical issues and availability of on-site experts.  

While AI has the potential to be a powerful tutor, a strong assistant and an equalizer in under-resourced classrooms, it can do so only if educators adopt tools with full information. School administrators should demand transparency, evidence and ethical safeguards from the companies that build them. 

District leaders should take these questions back to their board, curriculum team and legal counsel. Let them, not the slickness of a demo, drive the conversation. This can turn AI from a tempting shortcut into a responsible ally in education.

In the study, , two unnamed school districts carved out dedicated time for hundreds of elementary school students to work with a well-known AI reading tutor, either during class time or after school. Researchers followed about 350 students across two randomized controlled trials. All of the students were expected to log on for at least two 30-minute sessions a week.

They found that of the students assigned to work independently with the AI, just over 60% in the first district and 53% in the second ever logged on to the platform — at all.

Among all students, average weekly usage came to just over two minutes in District A and just over five minutes in District B.

Those who did log on averaged 13.2 minutes a week in District A and 25.8 minutes in District B, using the tutoring for just four to five weeks on average in an “intervention window” that ran from 14 to 31 weeks.

For Carly Robinson, the paper’s lead author and research director for the , the gap between access and use isn’t a shock. “As we’re talking about bringing AI tools into the classroom, the challenge isn’t just building good AI tools,” she said. “It’s getting students to use them and engage with them effectively.”&˛Ô˛ú˛őąč;

That’s going to take “intentional design” that appeals to both students and their teachers, who must choose whether to offer access.

“Having these tools available, even if they’re really good, doesn’t necessarily mean they’re going to get used if they’re not being embedded into kids’ learning experiences,” Robinson said in an interview.

But she was careful to note that the study didn’t draw conclusions about AI’s effectiveness, or the degree to which students were interested or uninterested in the bot, saying many factors could be at play. “This is not necessarily the students not engaging,” she said. In the two districts, the AI platform “was likely one of many tools available to teachers.”

For the study, researchers randomly assigned a group of students to work on the platform alongside a few classmates and a human tutor whose job was to support their engagement and motivation and to troubleshoot any problems students might encounter. In District B, the tutors were actually middle-school students who “had a free intervention block in their school day.” A typical session included a short check-in, 15 minutes on the platform and a few minutes of reflection.

Pairing students with a tutor worked, Robinson said — to a point. Usage increased by roughly one minute a week in District A and 4.4 minutes in District B. The number of stories students completed each week jumped 71% in District A and 80% in District B. 

What the human pairing didn’t do was move the needle on reading scores: Neither district saw a statistically significant improvement in end-of-year reading achievement. But Robinson said the study wasn’t primarily focused on that. Rather it was looking at the overall impact of adding a human into the equation, someone who provides “accountability, motivation and relationship building.”

Wednesday’s findings mirror recent ones from Khan Academy founder Sal Khan, who that the rollout of his in 2023 was “a non-event” for many students. “They just didn’t use it much.”

Khan said AI tutoring doesn’t necessarily make students motivated to learn, or to fill in gaps in their knowledge needed to ask questions.

The new data also raise an uncomfortable question for educators: Among students who used the platform on their own, those who logged on tended to be higher-achieving and less likely to receive special education services. So the students who stood to benefit most from extra reading practice were among the least likely to get it. 

Robinson said she sees that as a red flag for anyone considering AI tutoring as a quick fix for underserved students: “I think it should give us pause about treating AI tutoring as an equity solution.”

Alex Sarlin, founder of the newsletter and a veteran industry watcher, said the new study “shines a light on several of the most persistent challenges in ed tech implementation: low usage rates that don’t meet dosage recommendations, differential technology usage based on prior student achievement, leading to lower usage among the neediest students, and a faulty assumption that students will jump into new tools without structured guidance.” 

The researchers’ approach showcases a promising direction, he said, “as it is increasingly clear that providing access to tooling is not nearly enough to drive usage, let alone outcomes.”

Amanda Bickerstaff, co-founder and CEO of , which provides AI literacy training to teachers, said results like these aren’t all that surprising, given what we know about these tools.

All GenAI chatbots, she said, can make mistakes, lack important context about students and how they learn best, and can provide biased outputs. Her group has recommended keeping these tools out of the hands of students through second grade, “and only with significant human oversight and AI literacy training” for students in grades three through five.

“At this stage, there has been little evidence that GenAI chatbot tutors meaningfully impact learning outcomes for students,” she said, “or that they are developmentally appropriate for students in elementary schools.”

Robinson, the study’s lead author, said she sees the usage findings as part of a larger pattern playing out as schools adopt AI tools more broadly. Schools, she said, should consider offering students “different iterations of these things based on what they actually need — and that’s probably a more likely pathway to scale than just saying, ‘Let’s give everyone an AI tutor.’ ”  

Historically, personalized instruction has depended almost entirely on human teachers, with the teacher-student relationship central to the experience. But advances in technology — most recently in AI — have changed this dynamic, Robinson and her colleagues write. Now, personalized instruction exists on what they term “a spectrum of relational intensity,” from a consistent one-on-one human tutor to a computer platform that students navigate alone. 

AI tutors may approximate human interactions, Robinson said, but students may still benefit from the care and companionship that humans provide. Logging on and sticking with something that might prove to be difficult, she said, is easier with a human in the mix. “There is just this component of accountability that a human can provide, where it’s so easy to look away or check out of something when it gets hard when you’re dealing with a screen.”

A researcher at , a nonprofit focused on human connection in the age of AI, Lee asked: Why not?

In a word, the girl said: Instagram.

“If you try to dance at prom, you’re going to look stupid at some point,” Lee recalled her saying. Eventually someone will pull out a phone and you’ll end up on someone’s feed, seen by “the entire school” with mortifying results. Better just to play it safe. 

“Everybody just goes to prom to look cute,” the girl explained, “take a picture for the ‘gram, eat and leave.”

For Lee, who has spent years studying human belonging, that exchange unlocked an important, if unspoken, part of why AI holds such appeal. “We’ve created this set of conditions where young people don’t feel like they have permission to be their real, unfiltered selves,” she said in an interview. So they turn to AI, which is programmed to affirm them at every step.

from Lee and her colleagues offer this insight among others, painting a detailed portrait of how young people use AI and why. They surveyed 2,383 people ages 13 to 24 across the U.S. and found that for nearly half of them, AI has already reshaped their relationships in ways that are largely flying under the radar of parents, teachers and policymakers.

Among the findings:

• Just 15% of young people are in relationships with “personified AI” characters — but for about 45%, AI is already reshaping their real-life relationships;
• 53% of young people say they set clear boundaries with AI, using it alongside — not instead of — human support;
• 61% say parents rarely or never talk to them about AI, and 53% say the same about teachers;
• Youth from low-income households are three times less likely as others to engage with AI, but they report greater feeling: 21% feel lonely often or all the time, compared to 6% of high-income youth; 57% feel like a burden to others, compared to 42%; and only 34% feel a strong sense of belonging at school, compared to 62%.

For the study, researchers sorted respondents into four broad clusters. About 28% rarely or never use AI, often out of ethical reasons or just disinterest. The largest group, 39%, uses AI primarily as a practical tool. They turn to chatbots such as Claude, ChatGPT and Google’s Gemini for homework and research, while keeping clear boundaries between AI and their emotional lives. 

Another 18% use AI for personal and relational support, such as venting about a tough day, seeking relationship advice and processing emotions. And 15% engage with AI characters and personas in more intimate, companion-like ways.

Within the four groups, researchers found nine variations that challenge the conventional wisdom around AI use. For instance, among those who use AI for emotional support were two very different groups. Rithm calls them “Social Processors” and “Private Processors.” While they may look similar from the outside — both say they have lots of friends and use AI to work through their emotions — surveys found that the Social Processors use AI as just one tool among many. The Private Processors, by contrast, use it as a substitute for real human interactions because they feel they can’t bring problems to those around them.

That data point could hold the key to understanding problematic AI use, Lee and her colleagues said, challenging the idea that lonely teens with small social circles are most at risk of unhealthy AI dependence. The data suggest something else altogether, said Kashyap Rajesh, a rising junior at Cornell University who consulted on the report.

“The driver of risky AI use is not necessarily isolation,” he said. “It’s feeling like a burden [to others] — and that came through in the research.”&˛Ô˛ú˛őąč;

The number of friends a young person has, the size of their social circle, how busy they are, whether they’ve got family nearby and even their feelings of loneliness barely predict whether they’ll fall into dependent AI use, he said. “What actually predicts it is specific feelings: Feeling like a burden to others, feeling like you can’t be your real self, feeling like there’s no one to turn to.”

Julia Freeland Fisher, a researcher at the Clayton Christensen Institute who advised on the study, said that finding should help start a different kind of conversation around AI. “Burdening one another is building reciprocity, which is how we maintain the social contract, how we maintain social cohesion,” she said. That young people are increasingly bypassing this step should be alarming, she said.

“AI companions wouldn’t be nearly so disruptive to human connection if we had a sturdier social fabric,” said Fisher. “It’s the weakness of our social fabric that makes these [findings] so worrisome, not necessarily the technology itself.”

‘It just keeps feeling easier than the alternative’

For Lee, the finding on being a burden reframes so much of our understanding about young people’s relationship to AI. Virtually every survey respondent reported a specific “relational rupture” or crisis that made them turn to the technology. 

One young woman’s first question to a chatbot was, “I didn’t get asked to Homecoming — am I unlovable?” Another: “I got into a huge fight with my best friend, and I don’t want to tell anybody else because I don’t want them to take sides, so I needed to ask AI.”

“Story after story after story,” Lee recalled, “of a very singular, acute, discrete moment when they really had a moment of need and needed somewhere to put it.”

Rajesh, the Cornell student, said the data reveal a steady shift in which perhaps millions of young people are quietly moving from letting AI help with homework to asking it to mediate their emotional lives.

“They start off using it to help them write an essay, or help them prepare for their interview, or to study for an exam,” he said. “And they’re like, ‘OK, damn, this is really good, this is really helpful.’ And eventually their interactions escalate.”

The drift happens gradually, he said. AI helps draft an email or respond to a text. Next it’s helping to navigate a social situation. Before long it’s processing a breakup.

Rajesh, who’s studying information science and AI policy, said his own AI use crept up on him: He went from studying with Claude to creating personalized AI study guides to wondering if even attending class mattered. 

“I found that how many times I go to class and how actively I’m paying attention in class is actually not the biggest indicator of my understanding of the content or exam performance,” he said. “It’s actually just how much time I spend with Claude dissecting the lecture slides and building study guides that work for me.”

The report notes that because even productivity-focused platforms like ChatGPT, Gemini and Claude are engineered to interact with warmth and reassurance, what starts out as homework help or playful experimentation can evolve into a substitute for human interaction.

What adults get wrong

Alongside the findings on AI use, researchers found that how adults talk about AI is also potentially problematic: Their conversations are almost always about academic integrity — cheating, plagiarism, source citation — and rarely about relationships.

Rajesh said adults should be asking directly whether young people are using AI to process emotions, to rehearse hard conversations and to get support when they’re struggling. “Those are questions that signal to a young person that the adult knows this dimension exists and isn’t going to freak out about it — which is, I think, the prerequisite for any honest conversation happening at all.”

Michelle Culver, the Rithm Project’s founder and a co-author of the report, said young people tell researchers that when the topic is AI use, they’re “navigating it alone.” She suggested that adults approach the topic with “curiosity” rather than “judgment or shaming.” That could help both sides gain insight into each others’ struggles in the face of a technology that’s constantly challenging their reality.

In the same way that educators are worried that young people aren’t engaging in the “productive struggle” of learning academic content, Culver said, “We similarly worry that young people might offload the relational work to AI and become ill-equipped to handle the very messy human friction of real relationships.”

With many students and educators already using widely available artificial intelligence tools, state lawmakers and school districts are playing catch-up on AI policies.

In Maryland, for example, AI usage policies for K-12 schools are “all over the map,” Democratic state Sen. Katie Fry Hester said.

In some school districts, she said, AI use is encouraged, while in others it is restricted, or — a worst-case scenario for Hester — there is little to no policy guidance at all.

“What we heard repeatedly is that the teachers were feeling like they had to navigate artificial intelligence entirely on their own,” Hester said.

Hester said square one for lawmakers is AI literacy, which was the aim of new legislation that she sponsored and that was signed into law in May. It requires an AI coordinator in each school system, a statewide AI professional development for teachers and AI literacy to be a component of career readiness and computer science standards for K-12 students. It also requires the state Department of Education to provide certain guidance on AI.

Many other states have also been trying to create AI policies for schools. Lawmakers filed more than 134 bills across 31 states this year related to AI in education, focusing on data privacy, usage restriction in the classroom, literacy and training, according to MultiState, a government relations firm.

A survey by the Center for Democracy & Technology showed that (85%) reported using AI in their classroom during the 2024-25 school year, while 86% of students said they’d used AI for either personal or school-related reasons. But only about half of teachers and students reported that they received some training or information about AI from someone at their school, and few received training or information on risks of AI use.

A turning point for schools came with the rollout of ChatGPT in 2022, said Noelle Ellerson Ng, chief advocacy and governance officer for the School Superintendents Association. “AI was something that could not be gatekept,” said Ellerson Ng. “It was in the classroom the minute students were able to access it.”

Her association does not take positions on state AI bills or policies. But she said districts are trying to avoid knee-jerk, reactive policies such as New York City’s brief 2022 ban of ChatGPT because of fears about cheating.

Some states have made progress in laying the groundwork for AI policy in K-12.

Ohio has set a July 1 deadline for every school district, community school and STEM school to adopt an AI use policy. The state’s model policy recommends that districts address student and staff uses, privacy, ethical use, teacher-specific uses, vendor agreements, third-party AI tools and student assessments.

A new signed in March requires local school districts and charter schools to devise local policies for AI usage in K-12 schools, requires state standards for AI literacy and education training and ensures that no AI “replaces or eliminates a human teacher.”

enacted last month requires AI tools to be age-appropriate and requires teachers to review anything AI produces before using it in the classroom. It also allows parents to opt their children out of using AI tools. The law also directs the state education department to develop AI guidance and requires local school boards to set policies before the 2027-28 school year.

Yet even as schools are being sold on AI products by numerous vendors, there’s a growing skepticism about AI in classrooms. It follows a similar backlash about social media and digital technology’s academic and mental health effects on students, which has led to more states and districts putting in place bans and rethinking their reliance on laptops.

In the Center for Democracy & Technology survey, half of students said using AI in class made them feel less connected to their teachers, and 70% of teachers said they were concerned that students’ use of AI was preventing them from learning important skills.

Schools need to weigh the benefits of adopting AI tools in the classroom against their effect on student privacy, mental health and social skills, said Sue Thotz, director of outreach for Common Sense Media, a nonprofit advocacy group focused on technology and its effect on children and families.

Schools, Thotz said, may be the “only mandated safe space” where students can learn to use and access emerging technology. But she and other education experts believe districts need to increase scrutiny of products.

Globally, the market for AI products in K-12 schools was worth around $391.2 million in 2024, and could rise to more than $9 billion by 2034, , a market research company. That includes AI products for tutoring, personalized learning, automated grading, lesson planning and administrative tasks.

“When I talk about AI literacy, it’s not how to use AI. It’s understanding how AI is built,” said Thotz. “Why is it being created? Who’s profiting off of this?”

‘Giving a tool to children’

New York Assemblymember Robert Carroll said he uses artificial intelligence in his own work and sees its value. As someone who struggled with dyslexia as a child, he also thinks technology can help students with disabilities.

But he also wants to keep AI out of most K-8 classroom instruction. Students should learn basic subject matter first — in conjunction with critical thinking — and then later use the tools that can assist them, he said.

Carroll, a Democrat, has that would prohibit the use of most AI in K-8 classrooms, with exceptions for diagnostic testing and support for students with disabilities.

“It is imperative that all children gain strong foundational skills, especially in literacy and numeracy, and it seems that AI is uniquely positioned to possibly undermine that,” he said. “There’s a difference between giving a tool to adults and giving a tool to children who have yet to master skills.”

Rather than full bans, most bills seeking to restrict AI have opted to focus on age restrictions, parental opt-outs, oversight and bans on using AI to replace teachers.

This year, Florida’s would have included a statewide restriction on student access to AI instructional tools before sixth grade, with exceptions for use supervised by school personnel, English-learner translation support and disability accommodations. It overwhelmingly passed the Senate 37-1, but died in the House.

A adds computer science to the required public school curriculum, including AI and emerging technologies. Connecticut lawmakers in 2025 failed to pass aiming to stop AI from “replacing” public school educators.

Sophia Romee, the general manager of the GenAI Studio, an initiative studying how students and educators use generative AI at the College Board, the nonprofit that administers the Advanced Placement curriculum and SAT tests for high schools, said she is concerned that only that allow students to use generative AI have a formal policy governing its use.

The College Board’s research, Romee said, shows many students are worried about becoming too reliant on AI, and that adults need to give clearer guidance about where using AI tools for brainstorming, revising and tutoring crosses the ethical line into cheating.

“Students are far more self-aware about AI’s risks than headlines suggest.”

Like aviation in 1905

Jason Coley, director of the Center for Academic Innovation at Maria College in Albany, New York, said the policy debate needs to move beyond whether schools are “for” or “against” the use of AI.

“The better question is what kinds of AI use are supervised, age appropriate, transparent, and tied to real learning,” Coley said. Schools need guardrails around privacy, student data, bias, teacher training and equity of access, he said, but also permission to “experiment responsibly.”

Ellerson Ng, of the School Superintendents Association, said superintendents see AI as part of a larger umbrella of disruptive technologies in schools that has evolved from calculators to laptops to cellphones. The lesson, she said, is that overreactive policy rarely works. She also said schools should not cover AI in a separate policy, but as part of a broader technology policy.

“I don’t have a calculator policy. Why would I have an AI policy?” she said, describing how some district leaders think about the issue. “I have a technology policy.”

With past technologies such as cellphones and laptops, adults could often control when students had access, Ellerson Ng said. With AI apps and platforms, many students accessed the tools before teachers, principals or state officials were even aware of them.

That makes bans difficult, she said. Schools can block tools on school-owned devices and networks, but “you’re only one personal device away from social media and AI being in your schools.”

Justin Reich, an associate professor of digital media at MIT, said that uncertainty around AI should make policymakers cautious about declaring best practices too soon.

Reich said states are trying to regulate classroom AI at a moment when the field is still so unstable that “writing a guide for AI in 2026 is like writing a guide for aviation in 1905” before airlines, airports or even commercial flight.

“If you were to take any of the AI literacy documents, AI readiness documents, even the moratorium documents, and put them against a checklist,” said Reich, “there would be a lot of boxes in the ‘we’re making this up’ column and not a lot in the ‘we have evidence’ column.”

State lawmakers and school districts should be honest that they don’t know what they’re doing, are relying on limited expert information and that policy is subject to change with new information, Reich said.

“Lawmakers will need to be honest that what they propose now could be completely outdated in two years.”

is part of States Newsroom, a nonprofit news network supported by grants and a coalition of donors as a 501c(3) public charity. Stateline maintains editorial independence. Contact Editor Scott S. Greenberger for questions: info@stateline.org.

For many schools, it’s a race to keep up. Others are leading the pack. And some are unsure what to do with it.

Artificial intelligence is playing an increasingly prominent role in Hawaiʻi education, from the recent opening of the state’s first AI-focused charter school to the development of new coursework teaching students how to navigate rapidly changing technology. Teachers are also coming face-to-face with new technology, whether it’s using AI avatars to test students’ grasp of Mandarin Chinese vocabulary or confronting kids who are submitting assignments written by ChatGPT.

But there’s wide variation in how much teachers and students are willing to engage with the new technology. Civil Beat spoke with nearly two dozen educators, administrators and students across the state about their experiences with AI in schools and what they expect the future to look like.

A few years ago, the focus of AI in education was preventing students from cheating, said Michael Latham, president of Punahou School. Since then, he said, administrators and teachers at the OĘťahu private school have pivoted to teaching students how to use new tools responsibly, rather than banning them altogether. 

“We wanted our students to become really thoughtful and critical users,” Latham said, “to be able to understand these tools, but do so really with eyes wide open, aware of what some of the challenges would be.”&˛Ô˛ú˛őąč;

Public schools are piloting their own technology, including MagicSchool, a platform with AI tools like image generators and chatbots that teachers can tailor to their own lessons, said Winston Sakurai, educational administrative services director in the Department of Education’s Office of Curriculum and Instructional Design.  

While teachers receive training on how to use these platforms, Sakurai acknowledges that some are faster to embrace technology than others. It’s important for schools to find a balance between innovating with AI and meeting teachers where they’re at, he said. 

“We want people to embrace it,” Sakurai said, “but we cannot afford to burn people out with the adoption of it.”&˛Ô˛ú˛őąč;

Students also have varying levels of enthusiasm for the new technology. Some are emerging as statewide experts in the field, with one Punahou student developing resources to help teachers incorporate AI into their lessons. Others are more skeptical of the technology after they’ve seen their classmates cheat on assignments or receive incorrect information from AI chatbots.  

“I think personally, AI is a useful and dangerous tool,” said Charwin Irebaria, a junior at McKinley High School. “People should be cautious about what they try to get out of AI, because it can really affect our future.”&˛Ô˛ú˛őąč;

McKinley High School

In Cynthia Reves’ McKinley High School classroom, an assortment of items litter students’ desks: a bear-shaped pencil case, neon highlighters, a manila folder stuffed with papers. 

Laptops and iPads are conspicuously absent from students’ workspaces. Teens place their cellphones in mint green containers minutes after the bell rings and leave their devices there for the remainder of class. 

Like many teachers, Reves is questioning if her students are using AI on their assignments – and whether it’s appropriate for them to do so. In the past year, she’s tried to eliminate some of the uncertainty by doubling down on paper and pencil assignments and reducing students’ use of laptops and cellphones in her class. 

“What is dishonest use of AI?” Reves said. “That implies there’s an honest use of AI. What is that?” 

On a Friday morning in May, Reves ran her students through a series of short writing exercises to prepare them for their upcoming Advanced Placement English Language and Composition exam. Although the test would be online, Reves tasked her students with annotating passages and writing thesis statements by hand. 

“It feels like this huge tension,” Reves said. “I feel like I’m walking backwards, and yet there’s this AI thing that’s pulling forward.”

Reves went to school in the 1980s, when students used typewriters to complete assignments and Macintosh computers were the newest technology. Reves has constantly adapted to change in her 30 years in education, from teaching remotely to managing cellphones in class, but she’s worried about how the latest AI tools could impact students’ learning. 

In the past, Reves said, she could quickly identify when a student didn’t write a paper on their own. Just by plugging the opening sentences of a paper into Google, she said, she could tell when students found a copy of a similar assignment online and claimed it as their own. 

Now, Reves doubts herself. While she occasionally questions if students have written an essay on their own, they’ll deny using AI – and there’s no way to prove it, Reves said. Many AI detection tools have proven unreliable. 

Reves now confines most of her writing assignments to the classroom, rather than assigning essays as homework. If students want to finish a writing assignment during study hall, she said, she requires their teacher to sign off on the paper at the end of the period, affirming that the student completed the work on their own. 

But Reves also knows she can’t avoid AI forever. At the end of a recent English class, she assigned her students homework requiring them to read a short passage and answer questions. Once they completed the assignment, she said, they could run the writing prompts through AI to see how their answers compared to what the chatbot suggested.  

She wouldn’t have offered this option in August, Reves said. But her students have developed stronger writing skills and judgment throughout the year, and she believes they can now evaluate the AI responses’ critically.  

“I trust you,” Reves told her students.  

Mid-Pacific Institute 

Leslie Gleim’s preschool students at Mid-Pacific Institute are confident they can outsmart ChatGPT.

On a recent Wednesday morning, Gleim led a small group of 5-year-olds in a lively conversation about the observations they made on a recent field trip to HoĘťomaluhia Botanical Garden. She encouraged students to use their imagination to describe the clouds and explain how they came to life during their visit.  

“It was the happiest cloud!” one boy said. Another student speculated some of the clouds looked sad because the class hadn’t visited the garden for a few weeks. 

As the children chimed in, ChatGPT transcribed the discussion word for word on Gleim’s phone. Later, Gleim asked it to summarize the themes that emerged during the students’ conversation and compared the notes against her own observations.

The AI tool didn’t go unnoticed by the kids. 

“We seem smarter than Chat,” one student, Knox, said as Gleim checked the AI transcription on her phone. 

“Are you?” she asked. The group of kids erupted into cheers and enthusiastic nods. 

A few minutes later, Knox and his classmates put their theory to the test. One by one, they spoke into Gleim’s phone, asking ChatGPT abstract questions about their field trip to the garden and the personalities they gave the clouds during their discussion. 

After every question, Gleim asked her students to evaluate ChatGPT’s answers. Students were skeptical of the responses the AI generated – and Gleim was proud. 

“You guys won!” she said as students gave ChatGPT’s answers a resounding thumbs-down.

Using ChatGPT to record and transcribe her discussions with students has significantly saved her time over the past year, Gleim said. It also increases the stakes for students. When students know ChatGPT is listening to their conversation, she said, they’re more intentional with their words. 

But Gleim doesn’t want kids to think AI is the source of all answers. When she uses ChatGPT in front of her students, she said, she’s not afraid to point out its mistakes, using phrases like, “Chat didn’t get it.”&˛Ô˛ú˛őąč; 

“It’s not the gospel, but a tool,” she said. 

HawaiĘťi Technology Academy

Two years ago, Kingston Collman came to a sombering realization: AI was coming for his dream job.

Collman, a recent graduate from HawaiĘťi Technology Academy’s Waipahu campus, spent his early years of high school studying game development and planned to pursue a career in computer science. But as AI became more advanced, he realized projects taking him a year and a half to complete could be completed in minutes by new technology. 

“I was like, freshman year, AI is going to take over,” the 18-year-old said. “I don’t want to study for six years ahead and then be jobless.”&˛Ô˛ú˛őąč;

Collman pivoted to using AI to pursue his passion of producing social media content. For his end-of-year project, Collman developed an AI assistant to expedite the process of creating videos for social media. The assistant can generate scripts, suggest a list of video shots and assist with creating social media posts in a matter of minutes — a process that previously took him over an hour, Collman said.

The goal is to increase content creators’ efficiency, not to remove humans from social media altogether, Collman said, adding that viewers still want to see authentic content. 

It’s hard to tell what the future will look like for social media creators, however, and he’s worried more content could become AI-generated. Already, AI slop — low-quality videos created with AI — are becoming  on YouTube and other online platforms.Ěý

“You don’t know what’s real or not,” he said, “which is very dangerous.”&˛Ô˛ú˛őąč;

Some of Collman’s peers are equally skeptical of AI and how it could dominate fields traditionally untouched by technology. 

Adriana Hunt, another recent graduate from HawaiĘťi Technology Academy, is still reconciling how she can use AI to promote her artwork without compromising her creativity. For her end-of-year project, the 18-year-old asked ChatGPT to analyze her TikTok and Instagram art accounts and provide feedback on what types of posts were most likely to grow her online following. 

ChatGPT offered her advice she previously hadn’t considered, Hunt said, like adding more hashtags to her social media posts or posting more about her process of creating drawings. 

Hunt draws the line at using AI to create new art.  

When one of her teachers suggested using AI to generate ideas for her art or assist with her drawings, Hunt immediately rejected the idea. She uses art as a way to express herself and challenge her creativity — something that can’t be replicated by a few keystrokes on the computer, Hunt said. She’s also worried that relying on AI to generate art could set impossibly high standards.Ěý

“It’s never going to go away, because it’s now so woven into the fabric of everything,” Hunt said. “But we can learn how to rely on it less.”&˛Ô˛ú˛őąč;

KĹŤlia Academy

In James Morice’s data science class, sixth graders throw out phrases like “bimodal distributions” and “right-skewed graphs” with ease. 

At KĹŤlia Academy, a charter school in Kalihi, data science and coding are just as important to the curriculum as English and math. KĹŤlia is HawaiĘťi’s  middle and high school, opening in 2024 with ambitious plans to develop a seven-year program specializing in data science and coding.  

Before students can pursue careers in AI engineering and technology, Morice said, they have to learn the basics. 

On a recent Thursday morning, sixth graders at KĹŤlia were working together to examine graphs and input formulas in Google Sheets that could help them analyze a large dataset on the maximum speeds of roller coasters across the country.    

“What does every case represent?” Morice asked the class. “What are the attributes you see?”

The goal is to give students the expertise they need to understand how AI works so they can go on to develop new tools. 

Students still receive plenty of opportunities to work with new technology at KĹŤlia Academy, executive director Andy Gokce said. In English classes, Gokce said, students work with ChatGPT to receive feedback on their essays under the guidance of their teachers. Others have learned how to train AI models to distinguish between venomous and non-venomous snakes.

But Gokce doesn’t want his students to simply consume AI — he expects them to go on to lead the field and develop new technology as engineers or cybersecurity experts.Ěý

“We want them to know, inside and out, how it actually works,” Gokce said.  

Waiākea Intermediate School

Can students beat AI? 

It’s a challenge sixth graders were eager to tackle in Tyler Kojima’s world history class at Waiākea Intermediate School on the Big Island. 

On a recent Tuesday afternoon, Kojima tasked his students with defending a historical invention of their choice to an AI chatbot. He had programmed the chatbot to be skeptical of students’ inventions, which ranged from the Mesopotamian wheel to the woodblock printing technique. 

Kojima’s students could hold a conversation with the chatbot to convince it of the value of their invention, but they had to prove their point in eight messages or fewer. 

“You’re not trying to throw in the towel,” Kojima told his students. “You’re trying to beat it.”

For the next 30 minutes, students furiously typed on their laptops as they responded to detailed questions from the chatbots. 

“How do you know it’s more comfortable?” the chatbot challenged one student who was trying to sell it on the value of a new camel saddle. 

“Why should we trust ordinary citizens to make important decisions about our city?” it said in response to a pitch about Athenian democracy. 

It’s natural for people to be resistant to new innovations and change, Kojima told his sixth graders. When AI emerged a few years ago, he said, many adults were skeptical of the new technology — and so were his students.Ěý 

At the start of the year, Kojima said, many students believed AI was taboo. Most sixth graders don’t seem to use AI in their personal lives, Kojima said, and were wary about using it for the first time. 

Kojima is hoping to change students’ perceptions by introducing AI tools like chatbots and image generators that can make abstract topics in world history feel more concrete and exciting. He said he’s also careful to set clear limitations on how students can use the technology, instructing the chatbots not to give students immediate answers during class activities and occasionally rejecting the feedback kids receive on their essays from AI writing tools.  

“This tool is not going away,” Kojima recalled telling his students. “It’s up to you guys to know how to use it responsibly.”&˛Ô˛ú˛őąč;

Civil Beat’s education reporting is supported by a grant from Chamberlin Family Philanthropy.

“We’re in the middle of an industrial revolution that’s bigger than the dot.com revolution, and the world is not prepared for it,” Weingarten told ĂŰĚŇÓ°ĘÓ. “And our country’s leaders have a laissez-faire attitude about it. So I feel a huge responsibility to try and get it right.” 

Weingarten has proposed reshaping how U.S. public schools navigate AI in particular and technology more broadly, saying our kids are experiencing a crisis of attention and well-being — and that teachers are getting precious little guidance on how to help young people navigate these challenges.

Her proposal: Trim tech use, especially for younger kids, and teach all students how to think critically, communicate, collaborate and persist.“One of the worst things we’ve done in education was to call collaboration and communication ‘soft skills,’” she said, “because applied learning, problem solving, communication, collaboration, persistence — all of these — are the skills that any young adult is going to need in an AI world. In fact, these are the skills that are going to be much more competitive in an AI world.”

In a May 27 at the National Press Club in Washington, she proposed a near-ban on computer screens for students through second grade, including for assessments. She proposed banning student-facing AI in elementary schools, arguing that young children need to build foundational skills without algorithmic shortcuts. 

And she said that young people should not have access to “social companion” chatbots that simulate human relationships until age 16.

The speech makes Weingarten and AFT, the second-largest teachers union in the nation, new and potentially powerful supporters of a growing parent-powered movement to trim technology from U.S. classrooms, even as the union pushes to train thousands of teachers on how AI works. 

Weingarten proposed that schools redesign their offerings so that “active learning, including project-based, experiential and career-connected learning,” is the norm across all grade levels. She decried “drill-and-kill” rote instruction, saying that in an age when any fact is retrievable with a single prompt, the ability to apply knowledge, think critically, communicate and collaborate matters far more than memorization.

“To really prepare young people for complex challenges, our true goal is to have students who can work together and problem solve,” she said.

Weingarten noted that 31 states have now adopted some form of phone ban, and that several countries that were early adopters of education technology are pulling back. Sweden, she said, has returned to printed textbooks. Estonia, where research linked higher screen time in young children to weaker language skills, is calling for more human-to-human interaction. And Italy is re-emphasizing handwriting and traditional instruction.

Weingarten also called for establishing a rigorous new national safety and privacy standard for AI products sold to schools and creating an independently funded research consortium to study tech’s effects on children. And she proposed a new tax on Big Tech companies’ earnings to offset the environmental and societal costs of AI-driven disruption, including workers “being displaced by AI.”

In an interview Monday, Weingarten said AFT’s own $23 million AI academy, launched last year in New York City to help teachers understand and shape how AI enters their classrooms, exists in part to provide crucial guidance on how to understand the technology. Over the next five years, the National Academy for AI Instruction is expected to provide hands-on workshops for 400,000 educators, or one in 10 U.S. teachers, effectively reaching the more than 7.2 million students they teach. 

She said the institute’s mission and her new stance on tech aren’t incompatible.

“The AI Institute is really about teachers teaching teachers, and how the tech companies are not in control,” she said. “It is a people-first, safety-first focus.”

When she announced the academy in July, Weingarten said teachers face “huge challenges,” including navigating AI wisely, ethically and safely. “The question was whether we would be chasing it — or whether we would be trying to harness it.”

Nearly a year later, she said the institute now serves a crucial role in the absence of guidance from the Trump administration, which last week issued a U.S. Surgeon General’s urging families and schools to reduce children’s screen time. It suggested that schools limit school computers to computer labs, invest in physical textbooks and “prioritize pen-and-paper curricula, hands-on activities and social activities for all grade levels.”

In a media appearance last week, U.S. Education Secretary Linda McMahon said schools “need to embrace A.I., and to use it .”&˛Ô˛ú˛őąč;

Weingarten said it’s “crazy” that the U.S. Surgeon General’s office is offering more detailed recommendations than the Education Department. 

“When you actually have two-thirds of teachers in the United States having no idea how to use AI in schools, and when you have one-third saying there’s no formal guidance, and then you have the Education Secretary saying they should use it ‘appropriately,’ I mean, this is part of the problem,” she said. 

U.S. Education Department Press Secretary Savannah Newhouse said McMahon “has highlighted the many types of schools that are successfully and responsibly integrating AI in the classroom to help our nation’s students meet the challenges of today.”

Weingarten also took a swipe at Melania Trump’s recent tech-and-education event, in which the First Lady the White House alongside a humanoid robot to highlight the potential benefits of robots replacing teachers. The stunt, Weingarten said, “spoke volumes. So did the responses from teachers wondering how a robot was going to build trust with students or know when someone was having a bad day. There’s no algorithm for that. Students need their teachers — real human beings, not robots and not chatbots.”

Newhouse didn’t address Weingarten’s allegations about the administration’s leadership on AI, instead criticizing union priorities more broadly: “If there’s finally going to be an honest conversation about the damage done to American students, it should begin with the teachers unions’ enthusiastic support for a federal bureaucracy that has spent over $3 trillion only to watch student outcomes decline, along with their relentless push to keep schools shuttered during COVID,” Newhouse said. 

‘Kids are getting burned’

The effort to curb tech in schools comes on the heels of a similar one, led in large part by social psychologist Jonathan Haidt, to limit cellphone use in schools.

Weingarten on Monday said she has steeped herself in research on educational technology and artificial intelligence. But it wasn’t until she spoke to Haidt last summer about young people’s worsening that she knew she had to draw a line. 

“What really drove me was the issues around attention,” she said. 

Haidt, author of the best-selling 2024 book The Anxious Generation, has said short-form videos and other social media tools have decimated our kids’ ability to pay attention in school, resulting in fewer books read, poorer basic skills and worsening mental health. A more recent book, The Digital Delusion, by the educational neuroscientist Jared Cooney Horvath, argues that basic classroom technology has had a similar effect on skills.

In her speech, much of Weingarten’s criticism centered around increasingly widespread fears that our society is losing its way when it comes to young people’s technology use. She noted that more than half of 11-year-olds already carry smartphones, a figure that climbs to 95% among teenagers. Four in 10 teens report being online almost constantly, she said. “The pace of this tech revolution has been blisteringly fast, and kids are getting burned.”

She pointed to Haidt’s research linking heavy smartphone and social media use to rising rates of social isolation, anxiety and depression among young people, with academic consequences as well from the rollout of classroom technology. Scores on the National Assessment of Educational Progress, which had been climbing steadily, have in many cases worsened after widespread digital adoption. Weingarten acknowledged that correlation is not causation, but said the pattern, appearing consistently across states, grade levels and subjects, deserves serious attention.

She also pointed to research showing that 88% of teachers in a survey reported that their students’ attention spans were shrinking, which she attributed in part to the instant-rewards of online platforms such as TikTok and YouTube. Cognitive scientist work, she said, suggests students are not incapable of focusing, but are increasingly unwilling to do so when schoolwork feels dull by comparison to their online lives.

But she cautioned that she’s not anti-tech.

“I’m not calling for an AI ban or a Chromebook bonfire,” she said. “What I am calling for is getting the balance right to harness the benefits of technology while mitigating the harms. I’m wary of the dangers of AI, but it is here to stay. We need enforceable guardrails and help to cushion the disruption to people’s lives.” 

Alex Kotran, the founder and CEO of , said Weingarten is “right where it counts” about limiting AI for younger students but giving teachers access to the tools. “It’s about getting the balance right,” he said. “And I really don’t talk to anybody that believes that we shouldn’t have some sort of balance.”

Kotran said he’d recently spoken at an National Education Association meeting and saw that, like AFT, they’re focused on understanding AI. “There’s this almost-meme, ‘Oh, the unions are getting in the way of AI transformation, AI readiness,’ and I really disagree with that fundamentally. The unions have a very sophisticated understanding of what really matters here.”

Weingarten’s push to give teachers a better understanding of AI makes sense as well, he said. “When teachers feel like they are the main characters of the story of AI transformation, their willingness to really lean in and learn, it’s a lot more. You see a lot more buy-in.”

More broadly, Kotran said, supporting active learning, project-based and career-connected learning is “what all the smartest people in the field,” including CEOs and labor economists, are recommending. “What everybody’s basically saying is that the skills that matter now are people who can just get shit done, who can work independently and proactively on projects, who can create and build. And so it’s really, really important to hear a union actually naming that.”

On Monday, Weingarten said parents are leading the way on this issue — and that schools risk being caught between parents who opt their children out of classroom technology and those who want to keep it. “How does a teacher in kindergarten work in a classroom where half the kids opt out of screens and half the kids are on screens?”

The University of California at San Diego recently shook both higher education and K-12 when it a startling reality: Many incoming freshmen could not perform basic middle school math. 

The university was commendably specific about the causes: the COVID-19 pandemic and its educational disruptions, elimination of standardized testing, grade inflation and expanded admissions from under-resourced high schools. Together, these forces produced a class increasingly unprepared for quantitative rigor.

But here’s the paradox: These students looked highly successful on paper. Ninety-four percent had taken advanced math courses like calculus or statistics. They averaged a 3.7 grade-point average. One in four had a 4.0. 

So where did America’s K-12 school systems go wrong?

There isn’t a simple answer. But I would suggest a fundamental, largely unacknowledged problem driving these outcomes:

America’s schools are terrible at catching kids up.

A 2023 of nearly 3 million students across seven states found that those who started out behind academically rarely caught up. Students in the 25th percentile in third grade tended to remain in the bottom third through eighth grade and into high school. Low-income students, and Black and Hispanic kids,  barely moved at all. When the national ed nonprofit TNTP 28,000 schools nationwide, only 5% helped the average student catch up to grade level.

Kids who are behind stay behind. Kids who are far behind stay far behind. Call it the Catch-up Crisis.  

Why are schools so bad at catching kids up?

Because teachers are being asked to do the impossible.

In a classroom of 25 to 30 students, teachers must determine who is on grade level, who is behind (and why), how to modify instruction for each struggling child and how to extend learning for advanced students — all while delivering grade-level content.

Diagnostic exams, designed to give educators information on how students are progressing, are infrequent and often test different subject matter than what is used in the classroom. Intervention programs designed to catch kids up are purchased but poorly implemented. Students needing intensive help are sometimes segregated into programs with low expectations and weak outcomes.

In short, teachers rarely have the effective tools they need.

And in the absence of solutions, another practice creeps in: grade inflation (as evidenced by the incoming class at UC San Diego).

It’s hard to tell parents their child is behind. It’s harder still when the school cannot explain how it will help them catch up. So .

But when teachers have a roadmap for acceleration, honesty becomes possible. Poor grades become temporary markers on a path to growth, not permanent labels to be hidden.

There is reason for hope. TNTP’s study 1,400 schools where students consistently learned more than a year’s worth of material annually, enabling those who started behind to reach grade level. 

In other words, the Catch-up Crisis is reversible. But first. we need a bold, shared goal: that students who fall behind grade level will catch up to — or exceed — grade-level standards within two school years, and without fail by high school graduation.

Call it “On Track in 2.”

Governors and state education commissioners should adopt this goal publicly and report each year on how many students are behind, how many are catching up and how many are on track to do so.

Of course, setting the goal is easier than achieving it.  Doing that will mean tackling three big gaps for teachers: limited real-time insight into student learning, little evidence-based guidance on how to address specific learning gaps and minimal job-embedded coaching.

Before artificial intelligence, solving these at scale was nearly impossible. Students generate enormous amounts of work daily — assignments, quizzes, writing, projects. No human can analyze all of it for 25 students every day. But AI can surface patterns quickly and provide teachers with usable, digestible insights.

More profoundly, AI can help generate evidence-informed strategies for specific challenges.

Imagine a fifth grader who is struggling with fractions. His teacher knows he earned a C- on the last test but doesn’t know why or what to do to help. AI can analyze the student’s work in real time and discovers he tends to invert numerators and denominators; it draws on data from thousands of similar children to see what worked best to help those with the same misconceptions and recommends content for a 15-minute tutoring block for the teacher to review and revise.  

Of course, the teacher should always make the final call. But now she has a playbook to use as a starting point.

The same applies to feedback for teachers. High-quality coaching is rare because it is time-intensive and expensive, and the quality can vary without intensive oversight and training of coaches by the district. AI-supported , used responsibly, could provide timely, standards-aligned feedback on recorded lessons, supplementing human coaching rather than replacing it.

This is not science fiction. In the Bronx, Superintendents Cristine Vaughan and Harry Sherman have launched pilots designed to catch kids up in math using AI. They are partnering with organizations such as , and that provide smart and safe AI tools for teachers that assess students in real time,  identify what’s holding them back and recommend instruction that helps them get over the hump where they’re struggling. The pilots also include high-quality professional coaching for the teaching staff. Schools using these types of programs are seeing and increased student engagement.

These efforts are not yet the answer. And no technology should enter classrooms without strict vetting to ensure data privacy and security and to avoid adding to the problem with excessive screen time for students.  But they show the potential of smart and safe technology to give teachers new tools to catch kids up.

The evidence that the Catch-up Crisis is solvable is all around. It’s up to the adults who make and implement education policy to remove the barriers that are preventing America’s students from excelling.

Correction: The essay mischaracterized how many incoming freshmen at UC San Diego could not do middle school math. The correct number is 12.5%. Also, the name of one of the Bronx superintendents was misspelled. It’s Cristine Vaughan.

That era produced real gains, particularly in its early years. But over the past decade-plus, national academic progress has slowed, a trend that began before the pandemic and worsened afterward. 

The show that reading and math achievement remains below pre-pandemic levels nationally. Researchers and educators point to several overlapping challenges: rising student mental health needs, chronic absenteeism, widening opportunity gaps and the growing demands placed on schools both inside and outside the classroom.

To meet this moment, another incremental adjustment to the old reform playbook will not be enough. A new education revolution is needed, one that prepares students for a world shaped by artificial intelligence and puts their voices at the center.

AI offers the possibility of something truly transformative for students and educators, but only if it is approached differently than past waves of education technology.

Properly used, AI can help students research complex topics, test ideas and accelerate learning in ways that were previously impossible. It gives young people greater independence to create, question and problem-solve.

For teachers, AI can reduce routine administrative work and create more time for engaging, rigorous and deeply human learning experiences. Most importantly, it makes individualized learning more achievable at scale, particularly for students living in poverty, English learners and children with disabilities.

But education has been here before.

Schools once believed that putting a laptop in every student’s hands would transform learning. Devices alone, however, did not change instruction or improve outcomes.

The difference now is not access. It is adaptability. AI has the potential to reshape how individual students learn, but realizing that potential will require far more than adopting new software.

In the United States, education is deeply local. That creates room for innovation, but it also creates fragmentation. As schools begin adopting AI, districts across the country are developing different approaches, frameworks and expectations, often with little coordination or shared direction.

At a moment when the country should respond as it did to Sputnik — with a national call to action — there is a risk of disjointed efforts while other countries move forward with greater urgency and coherence.

AI will not replace human relationships or judgment, but it will reshape many aspects of work and learning. What is increasingly clear is that people who understand how to use this technology effectively will have a significant advantage over those who do not. 

That makes AI literacy essential. If every child deserves a fair shot at the American dream, then every student in every community must have the opportunity to develop these skills.

So where should that work begin?

Right now, many of the frameworks guiding AI in education are being written by adults: policymakers, technologists, researchers and commentators. Many are thoughtful. But most are still missing something critical:

The voices of students.

Today’s young people are digital natives in a world designed by digital immigrants, already navigating these tools with greater fluency than the adults around them.

Students will inherit the consequences of the decisions being made now. Shouldn’t they have a say in how AI shapes their education and future?

This summer, AASA, The School Superintendents Association, and Day of AI, a nonprofit initiative born out of MIT, are launching a that will bring together 50 school leaders representing every state to explore the future of AI in education.

Superintendents will participate in learning experiences on emerging AI innovations, breakthroughs,and the implications for schools alongside MIT experts, including , director of MIT RAISE and one of Time magazine’s 100 most influential people in AI.

But the school leaders will not come alone. Each will bring two students from their state.

Those students will participate in a parallel convening at the Edward M. Kennedy Institute for the United States Senate, where they will serve as “student senators” and work with Kennedy Institute staff to draft a National AI Policy for the responsible, productive and ethical use of artificial intelligence in public schools.

The student-developed policy will be shared with AASA’s nationwide network of more than 10,000 district leaders, giving students an opportunity to influence real-world conversations, decisions and guidelines on how AI is used in classrooms across the country.

Thirty years ago, education reform gained momentum when leaders across political and ideological lines rallied around the belief that schools could better prepare students for the future.

Today, education faces another inflection point.

If students are expected to live and work in an AI-driven future, they should help shape how that future is designed.

The next revolution in education should not be led by policymakers and pundits, but by students themselves.

The middle schoolers at Philly’s Marian Anderson Neighborhood Academy have a lot of questions about artificial intelligence.

They want to know how the government is using AI and what impact the technology has on the environment. They’re curious about how it’s being used for creativity, and whether it will be with us forever — or if it’s an economic bubble waiting to burst.

The sixth through eighth graders have been researching these topics and grappling with how it makes them feel about themselves, their education, and the world around them. On Friday, they presented their findings to their parents, teachers, and some state and local officials in their school cafeteria. Overall, they said there’s a lot they still don’t know.

Sixth grader Azizah Simmons said she’s weighed the pros and cons and she’s pretty confident that AI’s overall effect on our society is negative. If used correctly, Simmons said language models like ChatGPT could help kids her age improve their writing. More often than not, she said students use it to cheat on homework or cut corners on writing assignments.

But it’s the ubiquity of the technology that worries her most.

“You can’t really escape AI,” Simmons said.

Conversations about AI have permeated every aspect of education since the arrival of models like ChatGPT. Familiar debates about cheating have given way to Marketing pitches from companies promising “transformative” AI tools are now . In Philly, educators are working with students to build their own curriculum to and that can be embedded deep in the internal code.

And students say they feel like they have as much knowledge — or sometimes more — than the adults in their lives.

Sixth graders Thomas Mapp and Tyshaan Anderson’s research project focused on how video game designers use AI for level design, character creation, and visuals. Outside of school, they’ve been using AI to help them code games in Roblox and edit videos.

Anderson said he thinks the technology has helped kids like him experiment with creative fields like game design without needing to know the ins and outs of specific coding languages.

Marian Anderson Principal Nicole Patterson said she’s been inspired by her students’ civic inquiries and has learned a lot from them.

Patterson said she sees her school as a trailblazer in leading challenging conversations about AI. But she cautioned that “this is unfinished work.” She said students will continue their research and keep talking about these issues.

Marian Anderson computer science and technology teacher Trey Smith said the goal of Friday’s event was to help students and parents discuss how AI is now part of society, culture, politics, and everyday life, not just about how AI works.

“We’re all still trying to figure this out together,” Smith said. “For students to be in dialogue, not just with themselves and each other and me, but also with their families and with legislators and with school district officials and professors — I think it’s so important for them to learn together.”

That learning process can be tricky. Simmons said she ends up using AI involuntarily because search engines like Google now frontload AI overviews. That makes it difficult for young users to differentiate between what is a primary source link and what is AI generated.

“You use it without meaning to. It’s everywhere implanted in our lives,” Simmons said.

Chalkbeat is a nonprofit news site covering educational change in public schools. This story was originally published by Chalkbeat. Sign up for their newsletters at .

Yet just freely letting students use AI to generate ideas, explain difficult concepts and produce/revise writing ” upon which learning depends and . 

So I have been attempting the “third option” recommended by both the and the : teaching AI literacy.

This year my 10^th and 11^th grade English students used AI itself as a text to advance their critical thinking skills. We still read novels and short stories, still engaged in discussions and wrote essays, but AI was now a regular part of our work together.

As we read, we examined how large language models’ recycled novel “analyses” mis-read and oversimplified complex literature, producing distillations that often lacked nuance compared with the creative, discursive yet defensible readings that the students themselves generated. They learned to discern actual analysis from simplistic summaries, and to suspect the allure of AI’s instant “correct answers.”

As we engaged in literary discussions, we sometimes invited chatbots into the conversation; many students described these interactions as “bizarre” and “disjointed,” adequate for reviewing plot but too circular or directionless for genuinely provocative dialogue. ChatGPT’s sycophancy in particular tended to kill the necessary tension for true debate. One student “started purposely saying dumb things just to see how GPT would still find a way to say `great idea.’ It just felt so fake.”

As we wrote, we compared LLM-generated essays with human-generated ones, teasing out how AI’s “sophisticated-sounding” yet “generic” prose differed from the “messier” but ultimately, in the students’ judgment, more engaging language they themselves created. In a world where everyone has access to LLMs, these students were discovering the value of developing genuine voice. I hope at least some emerged thinking ChatGPT was best reserved for inter-office memos and letters to one’s utility company.

As we researched, we studied how AI search summaries — which users are now — don’t actually represent internet searches, but instead reflect word proximity within a static corpus of text, a corpus lacking access to paywalled scholarly research and therefore drawing disproportionately on unregulated chat forums. 

Students examined whether LLMs accurately reported their sources and to what extent AI drew from ideologically extreme sites. They saw how wording a query — e.g., “is abortion safe” vs. “is abortion murder” — could lead to politically-slanted results based on what the AI thought they wanted to see, and how sources often said something very different than AI summaries claimed they did. 

As we took and organized notes, students compared their manual note-taking process to the output of AI note-taking tools, learning how what we choose to include or exclude in summarizing notes, how we use emphasis and phrasing — did Africa under colonialism “fuel worldwide industrial production” or were African resources and peoples “exploited for the benefit of Western industrial profit” — create and propagate different narratives.

These narratives do not ; “what is ranked at the top” of AI searches “is ultimately influenced by the priorities of LLMs’ shareholders,” so we studied studying the politics of AI magnates like Sam Altman and Peter Thiel, learning how Gemini’s responses to political questions, and studying algorithmic bias (e.g., image generation requests for “doctor” returning mainly white males), all helped my students re-think their understanding that AI tools were neutral and simply utilitarian.

When we studied AI, we simultaneously studied neurological research about how humans, unlike LLMs, don’t just rely on pattern recognition, but also make intuitive leaps, and used Edward De Bono’s activities as practice. Students did something else that AI couldn’t: related classroom content to personal experiences. 

One multilingual student recalled attending a business meeting with her father where he faltered, because he “[knew] that someone who has the ability to speak English better [me] sat right next to him… ‘It makes me want to depend on you’ he told me, ‘when I’m totally capable of doing so by myself.’ He did much better after I left.” The student then made the leap to consider how, even if AI help is readily available, perhaps we gain something by refusing to rely on it.

When I abandoned AI bans, I instituted AI audits. Students had to demonstrate their thoughtful, detailed evaluation of each AI tool they used, including knowledge of how it operated, what they felt they gained and lost by using it, how they verified accuracy of information, and how they had not relinquished their own thinking. The students didn’t necessarily conclude “AI is always bad,” but they did see that using it always requires vigilance. Best of all, they didn’t have to take my moralizing word for any of this; they discovered it for themselves. 

Yes, I had to teach fewer novels in order to make room for AI literacy, but ultimately my job is not to teach novels; it’s to teach students. Their insights — how Grammarly’s “correcting” language altered integral parts of people’s unique voices, how personal evolution often comes from struggle and discomfort, how our desire for ease can hold us back from achieving our potential, how dangerous it is to invest authority in words just because they emerge from a machine — were equally valuable as any takeaway they gleaned from novels. And this time I knew those takeaways were theirs, not ChatGPT’s.

I teach an affluent population, but are with more economically and linguistically diverse learners. To be sure, my experience was often fraught. Some of my less-confident students never stopped considering LLMs’ “clear” and “well organized” writing superior to their own, and still hesitated to trust their own readings of literature over “the answers” ChatGPT offered. 

I struggle with asking students to critically evaluate AI while their own linguistic and analytic skills are still developing, but I also know I cannot create the conditions that allow teenagers to become master writers and thinkers before they are exposed to AI; they will soon arrive at my classroom having been using it since childhood. 

Post-pandemic suggests that, when teaching anything, we cannot wait for students operating well-below grade level to “catch up” before introducing higher order thinking skills; we have to figure out how to teach both simultaneously.   

That requires creativity, and creativity is what makes humans superior to AI, which can only regurgitate already-created ideas. Teachers excel at creativity; every day we come up with new ways to meet the ever-changing needs of our students, and right now AI literacy is one of those needs. 

that this training is crucial for keeping AI users — a population swiftly becoming synonymous with “humans beings” — from engaging in “cognitive surrender, marked by passive trust and uncritical evaluation of external information,” as opposed to “cognitive offloading, which involves strategic delegation of cognition during deliberation” when using AI.

about AI rendering English classes obsolete forget that the humanities are about studying what is human about us — including both our criticality and our adaptability.

Note: This is an abridged, non-scholarly version of a peer-reviewed article slated for publication in Issue 115.6 of NCTE’s .

Dacia Toll, co-CEO of and co-founder of , joins Michael Horn and Diane Tavenner to share how Coursemojo is using artificial intelligence to support students and teachers in English Language Arts. The conversation dives into how Coursemojo functions in real classrooms and the very human process it took to build the product itself.

Toll explains how she and her team started with the core curriculum — high quality instructional materials that build knowledge and vocabulary over time — in schools, then focused on how to ask students the right questions to gauge their understanding, give them the right feedback and then ask the right next question. They then figured out how to surface those insights for teachers in actionable ways. 

Listen below to hear about the professional development Coursemojo offers teachers and how AI makes it much easier to rapidly incorporate feedback and update the product, but, of course, with limitations.

Listen to the episode below. A full transcript follows.

Diane Tavenner: Hey, Michael.

Michael Horn: Hey, Diane. Good to see you and continue to crank on, uh, these, uh, AI tools that are starting to change what learning looks like in schools with you today.

Diane Tavenner: Today is gonna be a really fun one. I’m very excited to dive in with our guests today, but before we do that, this is our second time making this ask of our listeners, a second time in like seven seasons. And we never really thought this was important or quite frankly even thought about it, but it turns out it would be super helpful if you all could rate or review Class Disrupted wherever you’re listening to the podcast. And of course, please subscribe to it, and, and we’ve never asked you to do this because this is very much like a labor of love for us, but, but it turns out it kind of matters.

A little bit.

Michael Horn: Yeah, it’s absolutely true, Diane. And so a good way for other listeners to find out about it. And we of course get tons of private feedback from listeners, so we know you’re all out there. But, you know, if you can rate it, review it, subscribe it, you’ll help other people figure out as well what’s going on here. And while, as you said, this is a passion project for us, we do want it to matter and change the broader dialogue so people are having these conversations. And it turns out those, 5 stars, subscribe, beep, whatever it is. Those are a big deal, right, Diane?

Diane Tavenner: Yeah, they are. And then the last thing we’ll say about this is please keep telling us privately the things that we’re asking you to now say publicly, which is like what you like, who we should talk to, what’s working. We really do love the feedback and try to incorporate it every chance we get. And so please keep that, keep that coming.

Michael Horn: Keep it coming indeed. This is going to be a fun episode today though. It’s a friend of both of ours who, deep admiration for Dacia Toll, is our guest today. She’s a lifelong educator, school builder. She’s known for her work, obviously in 1999, founding principal of Amistad Academy in New Haven, Connecticut, dedicated to closing that achievement gap and then went on to found, of course, Achievement First. A network of many, I think 40-plus charter schools in the country, recognized nationwide as one of the highest performing school systems and so forth. And then in 2021, Dacia left Achievement First and soon after launched this company called Coursemojo, which we’re gonna get to talk about today. We’re gonna break it down.

It’s an AI-powered teaching assistant, but we’re gonna actually say what that in fact means, ‘cause it has very, very cool specific use cases that I think people are gonna enjoy learning about. And Dacia, thank you so much for being here. We’re thrilled to have you.

Dacia Toll: Very happy to be here with both of you guys.

Diane Tavenner: Yeah, super happy. I will say that over the years I have learned so much from Dacia on so many fronts. And so I’m really grateful to be here with you.

Dacia Toll: Right back at you, Diane.

Diane Tavenner: Yeah. Yeah.

Michael Horn: Well, I was gonna say, this is fun, right? We’ve all known each other and Dacia, you and Diane, you have something in common, which is you founded a school, then a CMO, and left a few years back and now both of you running edtech companies. We may come back to that. I should say you both raised venture funding. Like, there’s a lot of interesting things here. Listeners, of course, have a sense for why Diane made her job move, in my parlance. But why, why did you make yours? Like, what’s the founding story behind Coursemojo and the problem you were trying to solve by founding it?

Teaching, Outcomes, and Challenges

Dacia Toll: Yeah, so as you pointed out, I’ve been at this for a while now. I’ve spent a lot of time in my happy place, which is classrooms. Trying to figure out how to create the student experiences and the student outcomes, short-term, long-term, that we all want. And I do think we had a fair amount of, our students had a lot of success by at least the traditional measures. And then importantly, we always anchored in college graduation and launched into a career as part of what we were very focused on. But, it was hard. Like everything, you know, to really get a great teacher in every classroom with high-quality instructional materials and a strong classroom culture and relationships amongst kids and teachers and family and community connections. And then, I got inspired by Diane and tried to pull off project-based learning and expeditions and personal goals.

And sort of double down on, I don’t know, I don’t like calling them soft skills, but like the whole package, both as a parent, I now have teenage boys and as an educator.

Michael Horn: I always remember seeing your kid in the school, whereas you were making that transformation. It was so much fun to watch.

Dacia Toll: Yeah. Yeah. So yes, my own kids went to Achievement First schools. And so it’s just, it’s all very personal and I, I both believe as much as I did in the, in the possibility of success, but if we’re honest about it, it took so many things to line up to be successful. And I just, especially when AI emerged on the scene, I thought, wow, this, I’ve been frankly for most of that time an edtech skeptic. I think there have been lots of promises and it’s sort of overpromise, underdelivering is I think the pattern if you’re honest about most edtech. And AI felt different. Like we refer to it internally, I know others do as well as an electricity.

And you still have to build the light bulb or the power screwdriver or, you know, the tools that will leverage that electricity, but you are fundamentally working with something different now. And I found that inspiring. And so how, the problem we are focused on, it does feel like it’s been a lifelong effort, is, uh, reading achievement. Like, as you guys know, the NAEP scores in 8th grade are at the lowest level in 30 years. It does really feel like too many kids are falling off a cliff when it comes to basics of reading and writing. We could debate whether we think that’s still gonna matter in an AI-powered future. I do. And so it, and it feels really urgent.

And so we are, we have two big north star goals. We are trying to improve reading achievement. Specifically, we focus at the middle school level, although we’re now expanding to grades 3 through 10 next year. And then second, on teacher efficacy leading to teacher retention. Like we want more great people to stay in this profession. I think a lot of AI tools are trying to save teachers’ time. We certainly do that too, but we think we are in this profession because we wanna serve kids well, and that what will motivate you is if you feel like I’m doing a really good job at this thing that I care a lot about.

Michael Horn: So let’s dig in then, and just Coursemojo, obviously that tool, as you were saying, that you built using the electricity of AI to help solve that reading problem and boost teacher effectiveness. How do you describe what it does today in middle schools? What is Coursemojo?

Dacia Toll: So, there’s a student-facing side and a teacher-facing side. On the st— well, let me just say, take one quick step back before I dive in. You guys have been having so many fascinating conversations. Thank you for that. and people should like and subscribe and rate.

Michael Horn: Thank you.

Dacia Toll: Favorably, but what I, you know, I think it’s Bob Hughes and others who’ve talked about like multiple levels of AI in schools or models or paradigms. There, there is the one that I think is happening the most, which is finding efficiencies in small ways for teachers, whether that’s grading or on the operational side. And I’m, again, that’s great, but it’s sort of at the margins. And supplemental or not actually touching students at all. Then there’s this second model, which is more transformational when it comes to the teaching and learning experience. I think that’s at the moment where Coursemojo sits. And then there’s the third model, which is the AI-native schools. And I did listen to your episode with John Danner, and that I think is so.

AI’s Impact on Future Schools

Dacia Toll: And my personal belief as somebody who spent a lot of time in schools is that some people will want to and be able to make a leap to an AI-native school that’s an entirely different design, including in the, the goals that it’s trying to achieve with young people. But I think a lot of folks are gonna, hopefully make this transition into the model 2, work where it is meaningfully, it’s transformative in terms of what the experience is like, but it’s not a different universe. It’s like you’re still in schools, you’re still trying to do the core jobs, many of which I think are still important, although we could have that conversation if you wanted. So we’re going into the ELA classroom as it current — mostly as it currently exists, where you, especially in middle school, generally have content expert teachers who are trying to help kids improve their critical thinking about text and their writing skills, discourse, collaboration among students. And we start with the high-quality instructional materials. As you all know, it’s been one of the, I think, most positive steps forward to really have a quality curriculum that’s anchored in building knowledge, vocabulary, and reading skill over time. But that’s another one of those things that we all believe in, but is hard to execute effectively. And as a result, we have not universally seen the gains that we all believe we could. So we start there.

The first thing our team does is identify the hardest thinking part of every lesson. Which we know from the national research on these core curricula is often the part that gets a little watered down or skipped. Teachers run out of time, or frankly, they’re worried about the diversity of learners that exist. I think there’s on average a 5-grade-level span in a typical middle school classroom right now, so you can understand why teachers are anxious about giving the rigor of the text and task. So we identify the hardest thinking part of the lesson. Then for that part, not the entire thing, but for that 25-minute chunk. That’s where, for the kids, as a student said to me, it’s like the handout is talking to me. So it’s the same rich, wonderful text we’re already trying to read.

Adaptive Learning for Collaboration

Dacia Toll: It’s the same analysis questions we’re already supposed to be grappling with, but now kids are in partners or small groups, and we can come back to why that’s important to us, but they’re talking and then typing, and Mojo is like a learning buddy in that context. And what happens is Mojo figures out what the kid knows or doesn’t know about in response to that question, and then affirms, gives a little moment of metacognition if they need it, if they’re, if they’re struggling at like an insight, and then gives them the next just right question. And we could get back to why all three of those steps I think are important, but that’s something we worked on over time to make sure, we’re not trying to replace the teacher, but we’re trying to, as our great teachers have said, I can’t be in 27 places at once. So how can we get as close as possible in what a good teacher would do when a kid’s struggling, with a rich, meaningful question. So the kids are working. Meanwhile, the teacher has a live dashboard that shows every student’s level of understanding for every question. And so every class is wonderfully different, but In general, 85% of kids are humming with their partner and with Mojo, but the teacher knows right away the 15% of kids who are struggling for whatever reason, could be motivational, could be comprehension, and directs their effort and then sort of tees up for them what’s the gap between what the student’s current response and the criteria for success for that question. And so the teacher can, does, go around the room, conference one-on-one or with the entire small group, and push them forward as well. So we don’t think the AI’s gonna get every kid exactly where they need to be for deep understanding.

And in fact, we very meaningfully want the teacher to be focused strategically on what they can uniquely do well. So that happens for about 12 to 15 minutes that the kids are working on these close, generally close read questions, could be writing, and then the teacher goes over to, pushes a button and Mojo tees up the two biggest misconceptions in the class right now, and a suggested discussion question. So the teacher reviews them. We always want the teacher to make the choice about what’s the best use of time. Then the teacher pauses the class and facilitates a class discussion, not about every single question, but about the thing that is most holding kids back, often cuts across questions. And what we found is, I think it’s 85% of kids say they’re more likely to participate in class after having worked with Mojo. So on both the student side, they’re encouraged to participate. And then on the teacher’s side, they’re more confident because they kind of know what to go after.

And I think this just hits on another point. There are AI-powered learning experiences that are silent solo.

Diane Tavenner: Tons of them.

Dacia Toll: We, for a whole bunch of reasons, we’re talking about core Tier 1 instruction. We want it to be as beautiful as what these curriculum materials and teacher vision calls for, with more discourse, not less. And then again, there’s a lot we don’t know about the future for these young people, but we know we’re going to need our human skills more than ever. So that ability to work together, both in a full group setting and a partner setting, is really important. Anyway, and then finally it does end with an exit ticket, which is kids do independently, often writing. What’s different now is kids get, in every phase of this, kids get multiple rounds of feedback and then they revise their thinking and they revise their writing to make it better. And we know that also, like, if you don’t, I think about all the grading I did over the years and like, you just grade and the kids don’t.

You have to revise.

Diane Tavenner: Yeah.

Dacia Toll: So that’s inherent. And then there’s celebrations throughout. And Mojo also can make, it pulls exemplary student work, highlights kids who should be shouted out. So that’s the long whirlwind tour as to what it looks like.

Diane Tavenner: It’s awesome. Thank you for making it so concrete.

Michael Horn: Yes.

Reimagining Teaching with AI

Diane Tavenner: Literally taking us into a classroom about, and you know, anyone who’s been in a classroom, an English language arts classroom, what I taught certainly, like what you’re describing is what I aspired to do as a teacher, right? But I had to use my whiteboard or in the old days, my chalkboard. And I would, what I would call bumblebee around the classroom. Like you’re just trying to like bumblebee around so the kids are supposedly doing what they’re doing, but you’re not giving any feedback. You don’t actually know if they’re getting it or not, you know, like and so, it is making, well, in my words, it’s making the mere mortal be able to be the sort of superhuman teacher that we all want and imagine. And it’s just like my partner in doing that, right? I can be everywhere all at once and, and I, you know, have this brain working next to me and whatnot. What I know is that well, What I know from you in our previous conversations, and this is where I’d love to dig in and get a little bit nerdy right now, is I don’t— Michael and I keep pressing people to say like, what do you mean when you say by AI? You know, AI, because I think most people think that’s like, you know, logging into ChatGPT or Claude or Gemini and just asking a question. And you have gone through an extraordinary amount of work with AI to enable everything you just described. So will you take us sort of under the hood a little bit?

Talk about the type of work your team had to do to train the AI and get it to do the things you’re talking about and for you to feel good about it.

Dacia Toll: Yeah. This is something I both think AI is a game-changing electricity and I’m sort of frequently disappointed from a pedagogical perspective, even with the general large language models in terms of how well they evaluate student responses and how poorly generally they do at coming up with the next just right question. They’re so wired to give away the answer that it really is not the best pedagogical experience at this point. But what we do is, so first we start with the text and the questions, but then that’s nowhere near enough. It’s about how the AI, what the AI uses to evaluate the student response. And what they’re using is not just the knowledge of the unit and the unique text complexity and the lesson objectives for the day, but rather we have programmed in question-specific criteria. So they know if you’re analyzing this beautiful Langston Hughes poem and you’re answering this question that you need to— it’s maybe a vocabulary and context question— you need to know both what the word means and you need to know what it means in the context of this beautiful extended metaphor. and there’s not one right answer.

This is what’s complicated about the criteria of success, but there is a universe of good answers. There is a universe of typical kid confused answers, and that can be constantly refined. Like as kids teach us new insights into, you know, something like poetry, we can upgrade and do in real time. But so they’re criteria-specific questions, criteria, I’m sorry, question-specific criteria for success, which to my knowledge, I don’t know that anybody else is doing. Because it takes a lot of work on the front end.

Michael Horn: Now, I was gonna say, like, you have to go deep into these texts, right? I mean, just describe a little bit of that creation process also.

Dacia Toll: Yeah. And then there’s a whole other reading framework, but we can get to that. So yes, initially it required some of the best teachers we know on staff to do that kind of level of intellectual prep that you would do. Basically, what constitutes an exemplary answer? What are the transferable criteria of that great answer? So originally we human-powered it. Now we have so many examples of excellent human-powered criteria and have trained an AI authoring platform, which is internal-facing only, to give us a good rough draft. But we are still having humans do a gut check basically on, on— we could talk about what that looks like, but there are multiple steps. It’s like dominoes that the AI agent will tee up different things, and we still want those excellent teachers to go through and check.

Complexity of Reading Challenges

Dacia Toll: But so that’s what’s happening on the question-specific criteria. The other thing that we found is essential, which is why I really do believe for at least a long time, the specialized products are gonna outperform the general, like thin-layer products. And so we are very clearly geeking out on reading. It’s, as we know, reading is not math, and it’s not as simple. Like when a kid is struggling to understand a poem and to identify the central idea, it’s rarely that the problem is the skill of identifying the central idea. It’s almost always that there’s something else about the way that text complexity that is getting in the way, or it could be a fluency issue or background knowledge or a vocabulary issue, and we sometimes wanna say, oh, it’s a main idea problem, which then leads people to go outside the curriculum and do a bunch of main idea practice, which we know does not work. So like there are a set of ways in which we now have a whole reading framework that we’ve developed with, the good folks from Anet, Whitney Weldon, a whole bunch of reading experts, and it sort of honors the complexity of reading.

So Mojo’s looking at the criteria-specific success, it’s looking at the reading framework, and it’s trying to figure out what does this kid likely not understand about this? And that’s where there’s now this light bulb step that gives them like a little hint, and then it asks them the next just right question. And that is actually pulled from a bank of suggestions that is also, was initially human-authored, is now AI-authored specific to that question. So that’s not even tuning or training. It’s that the AI in real time is consulting with a set of resources and trying to pull the exact right instructional move for that kid. Would it be helpful to give an example or

Diane Tavenner: Yeah, well, because these have been created by expert teachers, vetted by expert teachers.And so like, I think the thing I want to, and Michael helped me here with the language, but like, there’s a lot of people, I would say the majority of people who are doing AI products are literally just putting like a wrapper around the language model right.

Michael Horn: With a bunch of instructions of guardrails around the context window.

Dacia Toll: It’s like prompt engineering, maybe.

Michael Horn: Yeah, not the training you just described.

Diane Tavenner: No, they haven’t literally taken expert professionals to work hand in hand with the AI to then produce this new experience, if you will, that brings the best of both of those. And so please do give us an example. I think it’s super helpful.

Dacia Toll: Well, I was in a classroom last week in Colorado, and the kids were analyzing a poem, and the question from the curriculum was, what role do lines 6 and 7 play in this poem? And these 7th graders were like, I really have almost no idea. So to the best of their ability, They are, they sort of, most of them tried to say what was going on in lines 6 and 7. And so what Mojo does is like, OK, affirm, first of all, good job, you know, in your own words explaining what’s happening in lines 6 and 7. But then the light bulb step comes. This question is actually asking you about the author’s craft and an intentional choice the author made to include these lines. Follow-up question: how would this poem be different without lines 6 and 7? That is like a — both the kid is like, oh, I didn’t even understand that this was about a choice the author was making, like author’s craft, and then I did — now I’m like guided into a process of actually trying to figure this out. And then if they struggle with that, Moja will say, well, what happens before and what happens after, you know, there’s like a set of additional questions that come out that what new teachers have told us, it’s one of the greatest compliments, is they do the Mojo activity before the kids so that they understand how to ask a scaffolded question without totally draining the rigor out of the thinking work that’s required, or how to give bite-sized feedback.

So that’s what we’re trying to do.

Diane Tavenner: Dacia, let’s stay here for a minute, as like a lifelong English language arts teacher. I wonder if I know some people will hear this and be like, who the heck cares if that author, like what their purpose was in those two lines? Like why, why are kids even learning this? Can’t we just teach them to read? So let’s spend a minute on how that transfers into the world and why that is so important and how, yeah, let’s start there. And then yeah.

Dacia Toll: Well, there’s so many layers to your question. And I’d love for you as a lifelong ELA teacher to offer your own point of view as well. But I think first we have the question, do kids in this new future still need to learn to read and write? And my, my strong conviction is yes, they do. Like, we’re going to be processing lots of information, but as we all know, there’s no firm line between listening, reading, and writing. It’s all the same cognitive process with each of them reinforcing the other. So learning to read is also a way in which, even if we believe AI is going to talk to us in the future, I think there’s still like the vocabulary and the, and the sort of way sentences get put together to effectively or ineffectively convey meaning. So that’s one. It’s like we could talk more about that. But I also think, I actually do believe we should be letting kids write more advanced pieces using AI.

That’s a whole separate other question. But if they haven’t learned to write themselves, I think that is a very dangerous place to start.

Diane Tavenner: Yeah.

Dacia Tol: So, that’s one. Second, it’s really just critical thinking.

Diane Tavenner: Well, that’s where I’m going. Like everyone’s talking about cognitive offloading and the lack of critical thinking and what you just literally put your finger on that activity is, oh. I can question or get curious about what an author’s intention was and why they did something. And that applies to every article. It applies to the —

Michael Horn: Well, it applies to the AI reading output you’re getting, right?

Diane Tavenner: Literally. Like, that is —

Dacia Toll: It applies to art. I mean, it applies to human interactions. Like, why is this person doing or saying what they’re doing in the way they are doing or saying it? And what does that reveal about them, their purpose? The message they’re trying to convey. Yeah, I think, I mean, just to, in defense of productive struggle, the brain, that’s the way we learn. Like if we don’t attend and focus and think and productively struggle, now there’s a zone in which that’s productive versus unproductive. But, and that’s part of what I think AI can help us get more kids like in their zone. But you have to, you have to remember. And forget and recall.

And these are the ways that the neural pathways get formed in our brain.

Diane Tavenner: Yeah. And I think your ability to like bring this into the classroom every day where it’s like, let’s just imagine now these young people every day in their class, they’re just doing this type of work over and over and over and over again. It is a muscle. You have to work it out. You have to build it. You have to practice it. You know, it will go away if you don’t use it. And as much as we’ve aspired for classrooms to look like this all day, every day, they, they for the most part don’t and haven’t.

And I’m not putting blame on anyone because it’s just so hard to do. And I think that’s why you call AI electricity, because it actually enables that, right?

Balancing Curriculum and Students

Dacia Toll: Well, I think so. Again, part of what we’re doing is going into the. Part of what you asked again, what set me on the journey, one teacher said to me, she said, am I supposed to teach the rigor of the curriculum or the kids in front of me? And the answer is both, but back to the whole thing being a little hard. Yeah. Like, she’s not wrong that bridging that, much less bridging that for 26 wonderfully unique individuals is very hard. And I think what too often we see happening in classrooms is the teacher, one of two things happens. Either they are so worried about kids struggling, and they’re not wrong, left to their own devices with that question, I watched it. A lot of kids out of the gate didn’t know what to do with that question. And so, the teacher holds it for the whole class and they ask the question and maybe, maybe 3 kids answer and then they move on.

And then the next question, same thing. And frankly, intellectual engagement is optional. For the other 24 kids in the room. And some of them are probably paying attention. Some of them, this is middle school, may not be or they’re going in and out of attention. The other option is I give it to them on a handout and kids are set to struggle and they put down the best answer they can. They put down the literal comprehension of those lines of poetry and then they move on.

Diane Tavenner: Yep.

Dacia Toll: The difference now is every kid is answering every single question. And intellectually grappling with it, and they’re getting real-time feedback that allows them to revise their thinking and their writing, and they get closer to encoding success before they move on. Cognitively, that’s a very different experience. Yeah.

Michael Horn: Yeah. I was gonna say the headline I’m taking away from this, right, is everyone’s worried right now about AI and cognitive offloading. You actually have used it to do the exact opposite, which is to make sure no one escapes this cognitive work and struggle, which is a significantly different use of use case of AI, but it’s significantly different use case of the classroom in a lot of average schools across the country right now. It leads into the question I have, which I’m just curious, like, is it hard for schools to figure out how to use Coursemojo? What does that on-ramp look like? Do they think of it as AI? Do they think of it as like an engaging digital learning activity? Are you, are you getting caught up in the backlash against edtech and the overpromising and underdelivering? Like, how is that all playing out?

Dacia Toll: Yeah. I personally lead a lot of teacher PD. I just love doing it and I learn a lot from it and it’s kind of fun to show up and we have a phenomenal partnership with Jackson, Mississippi and I was their leading teacher PD and they’re so nice. Like teachers are so nice. They come in, they, but the truth is they’re looking at me like, who is this lady? And it didn’t help that all they were told is like, come and learn about a new AI tool aligned to the curriculum. Like, That does not inspire confidence in the vast majority of teachers. And one of the things that’s wonderful is they actually have a significant number of experienced teachers. So it was actually teachers who had been teaching for a while and they start out a little side-eyed, like they’re nice, but they’re like, eh, skeptical.

And the first thing we do, like within the first 15 minutes of the PD, is they become students in a Mojo-powered classroom. And it’s often a text because it’s a curriculum that they’ve taught. And we start normally with, you know, we give, serve up a very challenging one. I remember in this case it was historical fiction about westward expansion, and they were like, “Ooh, the kids really struggled with this one.” And 5th grade. And then the teachers get into it and they realize how delightful it is. I mean, one thing I do really want to emphasize is we care a lot about joy and you should love reading. This should not be, yes, it’s cognitive and you have to productively struggle. But we are often like Mojo and the, and the way it’s organized, we’re taking delight in the text and the insights and the combination of affirmation.

And then when you get a partially correct or fully correct answer, Mojo has like just little emojis, like hundreds of them that sort of like your message the same way it would on social media or texting. And the kids love that. Like, oh, I got the little man on the surfboard or the muscle. And, and then if you get a 3 out of 3 on your writing after multiple rounds of revision, you get different gifts. And I saw a llama on a surfboard yesterday. And like, it’s the digital sticker. Like, we don’t overly gamify. They’ll never be playing Asteroid Blasters or whatever in reading class, but We believe in recognizing quality thinking and quality work, you know, the same way a great teacher does that praise.

And we make it easy for teachers to celebrate kids too. That’s even more meaningful. But the point is the teachers start out in this skeptical place and then they experience it and they experience the delight of my ability to identify who needs help and make my way over to them or facilitate the class discussion. And the other thing I would say is because they have this live dashboard teachers have told us they’re more comfortable letting kids work together in partners in small groups. Which I also think is something that I’m anxious about is there’s not enough of in classes in this, especially in this AI future. And because now they have, as a teacher said to me, eyes everywhere. They know immediately if that group in the back that’s gotten good at looking like they’re working is not actually working. So that’s important.

Improving Instructional Effectiveness

Dacia Toll: I think in terms of what’s hard, we also have the pleasure of working in New York City, and I was walking classrooms there with some coaches, and we were identifying that some of the small group work could be more effective. They were just happy it was happening, but they want— with it, it could be more effective, and that the full class discussion could also be more effective. Those are things they said their work— I mean, New York City has seen great gains with New York City Reads. But that’s the thing, that’s what they’re working on already. Like that’s the same way we’re all trying to improve effective instruction. Yeah. And I think the feeling was in the moment Mojo is nudging those behaviors to be enabling and nudging, but that there still is a level of teacher training and expertise that has to run alongside. So we’re asking ourselves, how could we support that even more in the context of the product with suggestions and sort of additional insight in real time.

But that’s really the hard part. I would say logging in is smooth and easy because of all single sign-ins. The dashboard is clear and intuitive. It’s colors that direct you where you need to go. So that’s not the hard part. It’s teaching and helping. We’re giving an alley-oop, but the teacher still has to execute some of those, those effectiveness moves.

Diane Tavenner: Dacia, along those lines, I think about this often, and I don’t want to give you heart palpitations because it gives me heart palpitations, but if you were back leading a network of schools again, like we, we used to, and I know neither of us, that’s not our chapter of life right now, but if you were there at this moment, what other opportunities besides this one, because clearly this is a need and you’re passionate about, like what else are you seeing in the world that you would be hopping on and wanting to bring into your schools and your network? What’s the sort of low-hanging fruit that you’d be going after?

Dacia Toll: Yeah, well, I think those are two different questions. Low-hanging fruit versus, I mean, what you inspired me to do, Diane, oh my gosh, more than a decade ago, was to create a whole new school model. And I do think back to this model 1, model 2, model 3. I mean, we’re all finding incredible efficiencies just by using AI and finance, the operations that there’s so many ways in which we should be doing that. And I am for saving teachers’ time. I’m most interested in the space I’m in, which is like, how do we improve the core teaching and learning? And I think those have to be in some level student-facing because, yeah. And I know there’s been some resistance to that and it has to be safe and, and pedagogically strong, but I would be trying to create a new model as well.

AI-Powered Learning Transformation

Dacia Toll: And I think that I don’t think it, you take it, you know, what we did inspired by you was first one school and then three schools. But I have the pleasure of being a part of a CIPRI fellowship, and they asked us to redesign, you know, to design a school, if we could, or a school model based on what we know now. And I do think the emphasis— I still believe knowledge matters. I still believe core skills matter. And the emphasis has shifted for me, like entrepreneurship, creativity, the human connection skills, leadership, judgment, ethics. And I think because we can potentially have the AI-powered learning experiences be so much more effective, I think that opens up more time for, I know a personal favorite of yours, project-based learning. I think there’s a way for AI to be embedded coaches in projects so that we, they were always so darn hard to pull off. But I think what’s exciting to me about this chapter, and the first thing we’re always talking to district leaders about and they’re talking to each other about is, Start with what’s your vision? What are your goals? What’s your vision? What are your values? And now increasingly AI can power a lot of that.

Now it takes this kind of specific design the way we’ve done it. I don’t mean to imply like you can go to ChatGPT and it will run your school for you. Like that is not how it’s gonna work. But if you decide we’re committed to project-based learning based on a knowledge graph, that can be powered now. And I’m hopeful that there’ll be more and more products that are trying to bring more and more of these experiences to life.

Michael Horn: Let me ask one last question. And Diane, I want you to answer this as well. So it’s a question for both of you. You didn’t wanna give yourself PTSD on running a network again. But, but, you know, we’ve talked about it, like you both founded schools, you both founded charter networks, you both had distinctive philosophies, enjoyed success. You both left, founded edtech companies. I’m just, I’d love to hear one reflection that you both have from being on the other side, if you will, on the edtech company side, or a company that provides to schools. You can view it either way.

One reflection that your school network founder self would have been surprised by at the time. Both of you, I’d love to hear your reflections.

Dacia Toll: Do you want to go first, Diane?

Diane Tavenner: I’m thinking, I’m thinking.

Dacia Toll: I will just say. There’s so many things, but what I was, it took a while for me to build new muscles, frankly, because initially what, what, particularly when you were running a system that got into a certain size, like as you said, we were 41 schools when I left, you had to set these big multi-year priorities and goals and you had to go after them in a sustained focus kind of way. And it was a lot about. Keeping this large organization aligned around the pursuit of those goals and cascading communication and systems to support this. And this is fast and iterative and responsive. And if we had tried to write a 2-year product roadmap, it would’ve been so painfully wrong. And so we start with vision and values. I’m not saying you don’t start with vision and values.

But then part of why I’m spending so much time in schools right now is watching and listening to kids and teachers and both what’s causing them friction in the moment and what their aspirations are. Like, what are they trying to get done they can’t get done? And then that literally dictates our product roadmap in the most wonderful way. And it’s really a code— we consider our school partners co-designers with us. Like it is that we’re running design input sessions with them about next future-facing things. Like it is fun and then the tech is just moving so fast.

Diane Tavenner: Yeah.

Dacia Toll: So like our ability to now have an agent that takes the do, all the work from today and creates your do now for you the next day in all of these things, like this is now that’s so different than it was 4 months ago.

Michael Horn: Wow.

Diane Tavenner: Wow. Yeah. That, that totally resonates with me, sort of the misalignment between how quickly well-run schools and systems can run and how fast you know, actually I should say how slowly that moves, even though you think it’s moving so fast as a school or network leader. And then this outside world is just going at a different rate and pace. And the co-building obviously resonates. I think what I think we’re in this moment in time right now where it’s really, I appreciate school and network leaders who are trying to think about streamlining and focusing and only partnering with so many people and not having a million platforms and how does it all integrate and whatnot. And I think the truth is that’s just not the reality yet. I think we should be striving to get there, but if we don’t kind of open ourselves up to what Dacia’s doing in this one little space and what other people— how do we actually ever pull together a model that has the best in class of everything versus just sort of this one generic, not good at anything approach? And so there’s a real tension there I’m feeling.

And on the tech founder side now, I’m like, how do I collaborate with, how do I build a community on our side that makes it more thoughtful and doable for the schools? And then on the school side, I would invite them to think about how do we, you know, sort of work more expansively right now while we collectively, you know, bring into the, our space all the possibility and then move to, I think, more coherent and elegant models over time. I’m not sure that was a-

Michael Horn: No it’s one of the hypotheses when we started Entangled Ventures back in 2015 or whatever it was, that and this was higher ed, but it was similar, which is like, you know, Arizona State University is getting pitches from like how many hundreds of companies every single day. They throw up their hands and they’re like, I won’t say the name of a textbook publisher, but we’ll go with them. We know every product is sort of mediocre, but like they’ve got everything. It’s just simpler. Right. And how do you bring forth a portfolio of, of best in class to an organization to simplify procurement and all that messiness?

Diane Tavenner: Yeah.

Michael Horn: But also give them confidence that they’re getting the best.

Dacia Toll: Well, I would say two things on that front. I don’t envy the tsunami of pitches that land in, I mean, I, it’s bad for me that the amount of AI-powered marketing now that lands in my inbox. And so this is, I think, a huge issue. And the more that others can step in and help school leaders make sense of all the different use cases, and again, what’s aligned to their vision and values. And then we haven’t talked yet about outcomes, but the North Star for us, as somebody who’s focused on ELA achievement, is, is ELA achievement improving? And we have actually a number of outcomes-based contracts aligned to that, which I know can be scary.

Michael Horn: So you’ve put your money, but you’ve put your money where your mouth is.

Dacia Toll: Yes.

Michael Horn: Yeah.

ELA Improvement with Mojo

Dacia Toll: And, and it’s, and we have seen, thankfully, huge improvements. I mean, as somebody who spent my entire life trying to improve ELA achievement and, and actually somewhat and successfully improving, it was always like 3 percentage points a year as reflected on the imperfect but consequential state tests. And we’ve seen across multiple partners, 6, 8, 10 percentage points on state tests in a single year. And, that’s not normal when it comes to— and, I think on some level, maybe it’s not surprising because we’re getting every kid to do the cognitive thinking. Those results are based on using Mojo 2 to 3 times per week. And, in general, we have much higher uptake and usage because it’s a core tier 1 because it’s everything else. But, I think it’s only Amira and Coursemojo on the reading side that have multiple independent efficacy studies. So, that’s not even just us, that’s ESSA Tier 2 research studies now that show that.

And I think people say, maybe this is why Diane and I, coming from the seats we were in, they said sometimes ed tech folks like, it’s too soon to evaluate impact. And I’m like, I just— kids are spending an entire year of their life in one of the most consequential classes of ELA, and we’re saying we can’t evaluate impact? Like, yes, the product is very different at the end of the year, thanks to all the feedback we’ve gotten along the way, but we still had this precious time with kids. Did it or did it not improve the core thing we’re trying to go after together? So, yeah.

Michael Horn: No, I think that’s very well said. All right.

Diane Tavenner: Yeah.

Michael Horn: Let’s wrap the conversation. We could clearly talk to you for a long time, but we’ve got one more segment before we do that.

Michael Horn: And before we let you go, Dacia, the fun segment we always have is something we’ve been reading, watching, listening to.

We try to get outside of education, but as Diane and I often note, we fail probably about half the time. So we’ll let you go wherever, wherever you go on this.

Dacia Toll: I am reading and thinking and listening to a lot of education-focused stuff. I would say one quick on the work thing, Lenny’s podcast has taught a tremendous amount about— I, as somebody who was making a transition into product and tech, I feel like it’s my weekly tutorial. But I also, we mentioned my teenage boys, so I try to spend some of that time with them and they’re getting me into anime. And so I just finished Death Note.

Michael Horn: That I didn’t have that on my checklist for you. OK.

Dacia Toll: No. Well, I, I have never watched anime before, but if it’s an opportunity for me to hang out with teenagers and then and you know, they’re pretty great stories of like heroes and even more, they’re kind of into the anti-hero, which then leads to a whole bunch of good conversations.

Diane Tavenner: That’s awesome. I love that. Anything that I can connect with my kiddos on is definitely something I will do. In this particular case today, I’m going to recommend a documentary film which is nominated for an Oscar, and this was my husband’s pitch to me, which was, I will say, not very compelling, which is there’s this new film out and it’s about death, and I really want to watch it. And then I think you’ll like it because apparently you also laugh at it. And I was like, wow, that’s not compelling at all. But it turns out that Come See Me in the Good Light is actually an extraordinary film.

Diane Tavenner: And it is the relationship of two poets who have these really interesting backstories and, and one of them is diagnosed with cancer. And I think he was wrong. I don’t think it’s about death. I think it’s about living and it’s a really beautiful film for this moment in time. So I recommend it.

Michael Horn: I was gonna say it sounded like a Shelly Kagan Yale, sort of course, the way you started to pitch it and then you changed that up on us. But I’m going to go a totally different direction because you’re outpacing me at the moment, Diane. But as I said in our last episode, it’s America’s 250th. And I’m going to give my brother a shameless plug on this one. I know we both read a lot on, you know, outside of our day jobs, but this one’s a little bit more personal because it’s my brother, Jonathan Horn. He’s been writing This Week in American History for the Free Press. It’s a weekly column, comes out on Wednesdays. It’s a fresh look at history that 

I’ve really enjoyed.

He names events happening around the country to commemorate, celebrate the 250th, including an event in Dorchester near me, which has helped us plan some outings with my kids, which has been super fun. But it was actually his piece on Thomas Jefferson a few weeks back that I highly recommend to all of our friends for its bigger messages and perspectives on the state of our union then, but also the state of our union now. And so highly recommend.

And I’ll just say, Dacia, huge thank you again. This was a fantastic conversation. We’re lucky you’re working on this. And for all of you, our listeners, keep the feedback coming both publicly and privately, and we’ll see you next time on Class Disrupted.

This episode is sponsored by LearnerStudio.

AI assistants like Claude, ChatGPT and Gemini won’t change these realities. But when you’re overwhelmed, they can help you streamline some of the most tedious aspects of your work. They can free up your energy for what only a human teacher can do. And AI assistants can actually push us to be more creative. They can help us overcome teaching ruts, nudging us to revitalize aspects of our teaching that are growing stale.  

AI assistants have arrived at a time when teachers need support to do their best work. In a national by the RAND Corporation, just 24% of teachers reported being satisfied with their total weekly hours worked, and 66% said their base salary was inadequate.

AI tools won’t make up for unfair compensation. But they can help us save time and create a better work/life balance. They can also help us do better work. 

A 60-second guide to prompting

The first step to making the most of AI is understanding how to use prompts. A prompt is a natural language instruction to an AI assistant. It doesn’t have to include technical or formal language. You don’t even have to use full sentences. 

Prompts are tool agnostic, so you can use them with whichever AI assistant you have access to. I recommend the free or paid versions of Claude, Gemini and ChatGPT, but you can also use free AI tools that run privately on your own laptop, like or . 

As the teacher, you guide an AI assistant like Claude or Gemini with relevant context. For prompts to work well they have to be detailed, including specifics and context. Generic prompts yield generic responses. 

It helps to iterate on AI output with follow-up prompts. Ask for more specificity or detail. Adapt the response for your students, don’t use it as is. Part of retaining your agency in the process is making sure you build on whatever outputs an assistant generates. You’re the director. It’s much like you were adopting an . The advantage, though, is that this material will be more tailored to your students and teaching approach.

Knowing how to use prompts effectively can mean the difference between AI that’s actually helpful and AI that’s gimmicky. The prompts below are designed to provide you with a creative boost. They each illustrate a practical way to use AI in support of thoughtful, pedagogically-sound teaching. You don’t need any special technical skills or subscriptions. You can copy, paste, and customize them to suit your subject matter.

Setting up a project

If you want to use prompts more efficiently, create a Claude or ChatGPT Project, or a . That’s a folder where you provide a summary of context about your students that the AI assistant can reference whenever you ask for support. You can also upload past materials, syllabi, lesson plans, curriculum guidelines, Common Core State Standards, or whatever else would be helpful context for the AI assistant. 

You can also provide detailed instructions in the project for how you’d like the AI to assist you. You’re training the AI assistant and teaching it your preferences. Once you set up a project,  you won’t have to repeatedly type in the same context. I set up projects for each of the classes and workshops I teach.

The Prompt Collection

The Bell Ringer 

Start class with a spark

The first five minutes of class set the tone for everything that follows. A short, well-designed opening activity can draw students in. Engaging openers are especially valuable on Mondays, after vacations or when you’re pivoting to a new topic. The challenge is coming up with fresh ones regularly. 

Goal: Generate a bunch of quick activities you can use at the start of a class session, adapted for your subject and your students. 

Prompts: “I teach [subject x] to students in [grade level x]. We’re studying [specific topic xyz. Be as detailed as possible about your subject and context. Include a sentence or series of phrases of context about your particular class and teaching style, or any special needs or context for your students. No need to make it formal].”

“Generate five bell-ringer activities I can adapt to open a [xx] minute class. Each should take no more than [x] minutes, require no materials, and either activate prior knowledge or help students reflect on what they’ve just learned. Include one that’s discussion-based, one that’s written and one that’s a game or visual/creative task. [Or adapt these examples to reflect your subject matter. For example, one of the options could be a logic puzzle or an artistic challenge].”

Prompt Example

The Real-World Hook 

Answer “Why does this matter?” before students even ask

When you’re juggling administrative meetings, multiple preps and paperwork, it can be hard to give extra attention to helping students relate to a given learning unit. This prompt helps you brainstorm connections to contemporary music, art, film, TV, cultural trends or other subjects of interest to students. 

Goal: To generate five ways to show your students how the topic you’re teaching is relevant to their lives, each with a two-sentence hook you can use to open discussion.

Prompt: “I’m about to begin a unit on [x topic] with [x grade level] students. [Provide a sentence of additional context and a few additional details about your students’ interests]. Generate five ways to connect this material to something students at [x] grade level may likely be able to relate to. This can include sports, the arts, social media trends, pop culture, music or other contemporary issues. For each connection, suggest a two-sentence hook I can adapt to help jumpstart a class discussion.”

Prompt Example

The Bad Example Generator 

Turn common mistakes into teachable moments

Showing students examples of common mistakes can help them avoid those pitfalls. But we can’t embarrass students by showing examples of their weakest work. Fortunately, AI assistants are excellent example generators. They can come up with nearly any kind of error you specify, saving you hours you might otherwise have spent creating intentionally bad work.

You can adapt this prompt to include any kind of error you want your students to avoid. These can include experimental design mishaps in science or mangled math formulas. If you’re teaching essay writing, showcase logical fallacies or ad hominem arguments. 

Here’s an example of a I generated with the help of an AI assistant.

Goal: Produce five realistic examples of a specific error type, unlabeled, so students can identify, discuss and learn from the flaws. 

Prompt: “I’m teaching [x subject/topic] to [grade level x] students. [Provide an additional sentence of specific context about your class, the learning goals you’re focusing on, and/or the lesson you’re preparing.] Generate five examples of paragraphs with [ad hominem arguments / circular reasoning / weak thesis statements / misleading use of statistics / or pick any other weakness] related to [x topic]. Make sure each example is realistic and plausible. These should be the kinds of errors students at this grade level might actually make. Don’t label what’s wrong. I’ll use these for a class activity where students identify and explain the flaws themselves. [You can also task the AI with annotating or explaining these errors to help you walk students methodically through these common flaws.]”

Prompt Example

The Scaffolding Prompt 

Make instructions clear for every student

Complex instructions often trip up students. Simplifying language can help, along with breaking guidance into smaller steps. This prompt helps you clarify instructions for an existing assignment, handout or any other activity. It’s particularly useful if you have students with learning differences or if your class has a wide range of readiness levels.

Goal: Reframe an existing handout or assignment so it’s clearer and more accessible, especially for students who need extra support. 

Prompt: “Here is a [handout / assignment / resource] I give students: [paste or upload the handout]. Help me reframe this for students who face [specific challenges or context that impact some of your students]. I particularly want this to be more accessible for students who need extra support. Break the instructions into smaller, numbered steps. Replace any abstract language with concrete, specific directions. Point out any parts I should clarify. Suggest a brief example for each major step and any illustrations or images that might help me make this more visually engaging. Maintain the academic expectations I have for the work. The goal is clarity, not simplification.”

Prompt Example

The Review Game Generator 

Create engaging questions efficiently for learning games

Coming up with a long list of review questions can take hours and designing multiple plausible wrong answers for every question can be exhausting. An AI assistant can help, quickly turning existing handouts, lesson plans or fact sheets into engaging questions. It can help you customize questions for your subject matter and student level. 

Goal: Generative 15 multiple-choice review questions, tiered by difficulty, formatted for whatever learning game you prefer. 

Prompt: “I’m finishing a unit on [x topic] with my [grade level x] students. I’m preparing an end of term review session, so I’m trying to come up with some good questions to help students practice [a particular skill or area of knowledge].  Generate 15 trivia questions based on the following key concepts: [list concepts or paste notes or upload a handout]. Suggest a series of multiple-choice questions, each with a correct answer and three plausible wrong answers. Vary the difficulty—five easy, five medium, five challenging. Flag the correct answer for each. Also suggest some true/false, fill-in-the-blank, and open-ended questions for variety.”

Prompt Example

The Fresh Angle Search 

Bring new life to familiar content

Some topics get stale, especially when you’ve taught them the same way for years. To liven up an old lesson, it can be helpful to gather new sources, examples, statistics, case studies or unexpected angles. 

Use , a free, AI-powered search engine that provides citations alongside its results. The links it provides ensure you have an evidence trail you can use to verify its responses and to dive deeper. Digging into Perplexity’s concise search summary is more efficient than sorting through hundreds of blue Google links.

Goal: Find five recent or unexpected real-world examples of a concept you’re teaching,  including perspectives from outside the U.S. and connections to students’ current interests. 

Prompt: “I teach [x topic] to [grade level x] students. [Provide additional context here about the topic or learning outcomes you’re focused on]. I’m looking for interesting material [or whatever other description you prefer] to make this subject more engaging for students. [Include any additional context about your students’ interests]. Find me five recent, unexpected or counterintuitive real-world examples of [x concept] that might surprise or intrigue students. Include also several real-world details to help add nuance for students who think they already understand the concept. And suggest several new analogies I can use for students who don’t yet understand this concept. Include international examples, and at least one that has an element of humor.”

Prompt Example

The Skeptical Student Prompt 

Prepare for the hardest questions before class starts

You never know what odd questions might arise when you teach a new topic. AI assistants can help by generating all sorts of potential questions. That prep can help you avoid unpleasant surprises in class, so you’re ready for nearly anything students might toss at you.

Goal: Generate 10 challenging questions a skeptical student might ask me about this lesson. 

Prompt: “Here is a [lesson plan / reading / concept] I’m teaching: [paste or upload material, mentioning the grade level and any other relevant context]. Give me a list of potential student questions about the relevance of this new topic and about real-world applications. Include also a mix of other unusual or surprising questions curious students might ask. If these high school students doubt this material is relevant, what might they ask, and what aspects in particular might they question. Generate 10 challenging questions students might ask. Include questions that challenge the relevance of the topic, the reliability of my sources and the assumptions behind my explanations.”

Prompt Example

The Blind Spot Audit 

Find your own blind spots before students do

During a typical week, we don’t always have time to trade peer feedback on lesson plans or syllabi. But we can still benefit from getting input on our materials. AI assistants can critically evaluate your materials for clarity, accessibility, inclusivity or other blind spots. You always have the option of ignoring the observations. I find that many of the weaknesses the AI assistant points out are ones that benefit from a fix.   

Goal: Identify specific places in your lesson plan or syllabus where I might have an unconscious bias, where my instructions may be unclear, my examples may not reflect student diversity or my assessment criteria might be confusing. Or point out unnecessary jargon.

Prompt: “Here is my [lesson plan / syllabus / unit overview]: [paste or upload document]. Take the perspective of a critic with expertise in inclusive pedagogy and student-centered design. Identify parts of my plan that may not work for someone with physical differences such as a vision, hearing or mobility impairment. Also point out places where an unconscious bias might be influencing the way I’m presenting this topic. Point out places where examples or explanations I’ve included might not make sense to my diverse students. Show me places where my assessment criteria could be made more clear. Note any other sections of the material that might not be inclusive, accessible or relatable for students. Be direct. Include the location of each issue so I can explore potential fixes. I want specific critique, not general praise, and I want you to explain each observation in detail.”

Prompt Example

The Differentiation Prompt 

Adapt one assignment for three distinct student levels without tripling your prep time

In many classrooms, students arrive at varying levels of readiness. Creating three versions of the same material is one of those things that turns a 40-hour week into a 53-hour one. Tasking an AI assistant with suggesting adaptations of your material ensures that your newly differentiated materials will remain anchored in your own ideas and teaching goals. 

Goal: Produce two alternative versions of an existing assignment: one with additional scaffolding, and one with stretch challenges for advanced students.

Prompt: “Here is an [assignment / assessment] I give students: [paste or upload material]. Generate two versions of this: one for students who need additional scaffolding and more explicit guidance, and one that adds stretch challenges for advanced students. Preserve the core learning objectives. Summarize the suggested changes and explain their rationale, so I can decide how to adapt these alternatives for my students.”

Prompt Example

The Rubric Builder 

Help students understand how you’ll assess them.

A well-designed rubric does two things: it clarifies your expectations before students start working, and it gives them a roadmap for revising. Developing rubrics from scratch is tedious. It requires formatting small batches of text into boxes in complex tables. This prompt generates a structured first draft in table format. You can then refine it before sharing it with students. To start, specify the elements of the student work you’ll be evaluating, and describe your criteria. 

You don’t have to use full sentences or formal language. Just describe what constitutes excellence for this assignment, what satisfactory work looks like, and what evidence signals to you that a student may need more skill practice. Developing these rubrics with AI assistance is an iterative process. Revise initial outputs by adding your own details and refinements.

Goal: Generate a rubric with three performance levels and five criteria you’ve specified, written in specific, concrete language, without vague phrases like “good use of sources.”

Prompt: “I’m assigning [describe assignment] to [grade level x] students. [Provide any additional relevant context]. Generate a rubric with three performance levels: Excellent, Proficient and Developing. Include five criteria relevant to this assignment: [list criteria, e.g., argument clarity, use of evidence, originality, structure, mechanics]. For each criterion and each level, write two specific sentences describing what that performance actually looks like. Avoid vague language like ‘good use of sources.’ Be concrete. Put this rubric into a table, then await my input for potential edits”

Prompt Example

The Case Study Collaborator 

Generate fictional scenarios to spice up discussions

Case studies help spark lively discussions. They’re useful whether you’re introducing students to ethical questions or trying to help students relate to a historical situation. They can also be useful for bringing a business decision or a scientific discovery to life.  Creating cases from scratch can be exhausting. So this prompt helps you build fictional but realistic scenarios customized to your subject matter and student context. 

Goal: Create a fictional case study to illustrate a tension relevant to your subject, set in a context students can relate to, ending with three discussion questions.

Prompt: “I teach [x subject] to [grade level x] students. [Provide an additional sentence of context or specifics to ensure the case studies are relevant and useful.] We’re exploring [x concept or issue. Include as much detail as possible about what and how you’re approaching the topic and your learning goals]. Create a fictional but realistic case study involving [type of character, institution, or situation relevant to your subject] that illustrates the tension between [value A] and [value B]. Set it in [context relevant to your students—a school, a local community, a specific industry]. The scenario should be complex enough that reasonable people could disagree about the right response. End with three discussion questions that I can adapt to push students to apply the concepts we’ve been studying.”

Prompt Example

Disclosure: Two kinds of prompts appear in this piece. I developed the templates with brackets based on my teaching experience. The filled-in examples showing how teachers might customize each template were drafted with help from Claude, an AI assistant. Using AI to help generate these examples let me stress-test and customize each template across different subjects and grade levels and confirm that the prompts produce useful results. I reviewed and edited every example.

Traditional end-of-year exams were intended primarily for accountability purposes. While useful, they were never built to provide a real-time understanding of student growth or inform instruction. But despite technological advances, they’ve largely defined state assessment systems for more than two decades.

First enshrined in federal policy through The No Child Left Behind Act, these rear-facing tests offer a snapshot of performance after learning has occurred, giving educators little opportunity to adjust teaching or address gaps in real time. Preparation for these exams eats into valuable classroom learning time, and the cost of purchasing and administering them represents a major line item on state . At times, their shortcomings have for high standards.

Back when these exams were developed, schools had few options for addressing the staggering and unacceptable achievement gaps facing students. And, in places like Massachusetts, where I helped craft and implement education reforms three decades ago, the balanced carrot-and-stick approach did lead to improvements for some time. 

The Massachusetts Comprehensive Assessment System, which I helped steward, was one of the nation’s first statewide testing programs to quantify academic performance. The MCAS was critical in driving achievement across the state, and Massachusetts students scored at the highest levels in math and English on the . Our state was named a national model by the U.S. Department of Education, and the MCAS served as a blueprint for other states. 

This was in the mid-1990s, and statewide assessments — albeit clunky, expensive and time-consuming — were as revolutionary as the car phones being installed in center consoles. For the first time, states could provide parents, educators and policymakers with objective data for identifying failing schools and specific student needs.

But while mobile phones and Blackberries have long since replaced car phones, the world of education has clung to the same, outdated approach to testing. What’s worse, rather than replace the exams, states and districts — faced with both an appetite for data and the limitations of end-of-year testing — layered in more assessments in the form of interim exams, diagnostic tests and progress-monitoring tools. The goal was to generate more information that could be used immediately to inform instruction. The result, too often, has simply been more testing. At the risk of extending my car phone analogy too far, they bought and used fancy new iPhones while still paying for and insisting on using that old car phone when making calls on the road. 

Today, states and schools have an opportunity to do this differently. AI-enabled assessments, for example, can listen as students read aloud during normal classroom practice, continuously gauging fluency, accuracy and comprehension in real time — so the assessment is happening while learning is happening, not taking its place. This, in turn, can equip educators with immediate feedback to make lessons more effective and ensure that what’s being tested aligns with state standards.

In the past, efforts to adopt new measures have been thwarted by federal regulatory constraints and limited technological capabilities. But many of those longstanding barriers to innovation are disappearing. Today, artificial intelligence makes the need for repositories of test questions a vestige of the past, instead automatically adjusting difficulty levels based on a student’s answer.  And speech recognition allows a once unimaginable capture of early literacy progress — even for children who cannot yet take a test. 

Perhaps most importantly, the current administration has made clear that states now have the flexibility to explore new options. The Every Student Succeeds Act still requires annual testing, but it also permits states to choose between administering a large-scale, end-of-year exam or multiple interim tests that combine their results into a single score.

states are already using this flexibility. recently adopted a model that assesses students at the beginning, middle and end of the school year to measure progress. recently received federal approval to pilot a similar model that gives students multiple opportunities to demonstrate mastery and provides educators with frequent insights about where their students might need support. are exploring similar approaches. 

With modern technology, these models can go even further. Computer-based assessments can now capture the voice of a student reading aloud, solving a math problem or completing a writing task to generate meaningful data without pausing learning for a separate exam. The result is not simply more information, but more useful information — timely, aligned to core math and reading standards, and capable of providing timely feedback about where students are succeeding and falling behind.

These assessment models are already operating in classrooms across the country, and they can produce reliable measures of growth while correlating strongly with traditional benchmarks. More importantly, they give teachers insight when it can still change outcomes for students. That distinction matters — especially now.

National data show troubling declines in math and reading performance. Achievement gaps are . High school students are performing below levels seen two decades ago. In this environment, diagnosing learning loss in June is insufficient. Educators and students need systems that accelerate learning in October.

An assessment should not be an autopsy. It should be a compass.