Sign up for Chalkbeat’s free weekly newsletter to keep up with how education is changing across the U.S.

Paige Wyatt heard plenty from administrators about using AI to generate lesson plans at the school where she taught high school English.

Her students used AI to cheat so frequently she resorted to spending her own money on detection tools. Yet she never got help with what she really wanted: how to effectively catch and stop the cheaters, some of whom were very clearly just copying and pasting from Google’s AI summaries.

Teachers like Wyatt consistently identify cheating as a major concern in the AI era, and 70% of teens report using AI for schoolwork, a Common Sense Media survey published last month found. But that concern is getting little recognition from states. While at least 13 states have passed laws about AI in education, the laws most often direct state agencies and districts to create their own AI guidance, meaning lawmakers have frequently absolved themselves from making tough decisions.

RSVP for Chalkbeat's virtual event on the two tech debates shaping schools this year: AI and screen time.

Meanwhile, roughly three out of four states have published information to help guide AI use in education. But a Chalkbeat review found these documents tend to skirt the issue of cheating, offering little in the way of concrete steps to halt or deter students from offloading work to AI. Instead, they dedicate more real estate to AI literacy and ways teachers can use AI for efficiency.

All of that beating around the bush has left educators and districts to their own devices to come up with ways to deter and catch AI-enabled cheaters. It’s become an exhausting endeavor that hasn’t turned up any definitive solutions, educators and researchers said.

There’s no surefire way to detect AI-generated content. Nor is there an agreed-upon definition for cheating: Can students use AI to outline an essay but not write it? Can they use it to copy edit? Some states have crafted acceptable use policies which make some, but not all, of these distinctions.

Wyatt quit her teaching job last year. She didn’t leave solely due to AI, but its impact on her work was a factor.

“We really need real consequences,” she said. “We need workshops for teachers to detect cheating. We need workshops for kids.”

Most of the nation’s 10 largest school districts are developing or have developed more specific policies to try to prevent using generative AI to cheat. But it hasn’t always been easy: New York City delayed the rollout of its AI guidance for months amid fierce backlash, before announcing a new policy this week. Half of those districts explicitly discourage using AI detection tools.

Many of the official responses might underscore the urgency of the problem without helping to address it, which could only increase teachers’ anxiety.

“I think the situation teachers are in feels bad, and feels like they ought to have a solution now that they don’t have,” said Justin Reich, a professor at the Massachusetts Institute of Technology and the host of a podcast about how AI is shaping K-12 education.

Letting AI ‘run amok’ could ruin student learning

Historically, state policymakers haven’t rushed to create laws governing new technology and education.

States didn’t really start to adopt computer science requirements until the 2010s. And while the iPhone launched a smartphone revolution in 2007, the first state restrictions on cell phone use in schools didn’t come until 16 years later, in 2023.

States may be moving faster when it comes to mainstream generative AI, which took off with the advent of ChatGPT in late 2022.

“Folks can’t just hope this will pass. This is a major change, and students are already using it,” said John King, former U.S. education secretary and current chancellor of the State University of New York system. “This isn’t tomorrow’s problem. This is today’s problem.”

The District of Columbia and at least 37 states have issued recommendations for public schools about AI. Their documents range from short webpages to Illinois’ 400-page behemoth. Many define different types of AI, stress that any AI use should still be led by humans, and link to outside resources — some of which have been developed with the support of big AI companies.

But they often devote just a page or a few paragraphs to academic integrity.The documents also seem light on actual guidance and don’t include many citations, Reich said. That’s because there’s little evidence to show how effective AI tools are in the classroom, or what policies would effectively stop cheating.

“If you just let this run amok in secondary schools, all kinds of learning activities will stop,” he said. “We do not know what to do. There’s no evidence-based policy. There’s no basis for development and professional development. Like we’re just making stuff up.”

More than a third of states urge schools not to rely solely on tools that promise to detect the use of AI, citing studies showing they are not effective and incorrectly flag genuine student writing as AI.

But many states then offer little else other than acceptable use policies. In California’s case, the state offers a list of questions for districts to mull when making policies around integrity in addition to an acceptable use policy.

In the absence of hard and fast rules, teachers have reported requiring students to complete more graded work in class, using pencil and paper.

Christopher Harris, the director of the school library system for the Genesee Valley BOCES, a regional district in western New York, wants educators to go further than that and reconsider how they measure learning. Writing may be less of a valid tool to assess if students are learning, now that AI “can suddenly write anything.”

He said students could submit an oral assignment, or teachers could focus on the learning process more than the product by asking students to explain how they respond to feedback when developing a project or presentation.

New York is one of the 13 states that hasn’t published AI resources for educators. Harris’ library system has been developing resources about AI for the state’s educators using grant money.

“I really wanted to help teachers understand they can’t control what kids do with AI,” he said. “The only control they have is their agency in developing assignments.”

Adopting AI policies not straightforward for school districts

Sara Brescia thought she had a no-brainer policy proposal for her fellow Manassas City Public Schools board members: adding AI misuse to the district’s code of student conduct.

The policy she proposed for the district in northern Virginia — a hotbed for the data centers AI relies on — would prohibit students from trying to pass off AI-generated content as their own. It specified that AI could only be used with a teacher’s permission, and required students to cite AI use. Students would be punished, up to and including with expulsion, for violating any of those provisions.

But other board members voted her idea down in late July. Lisa Stevens, another school board member who is also a teacher according to her LinkedIn page, worried the policy would add a burden onto teachers without giving them any tools to identify AI-generated work.

Brescia is skeptical about the value of those tools.

“We do not have a comprehensive approach,’’ she said. “We have the student code of conduct that says nothing.”

Wyatt said when she left the classroom, students were only becoming more fluent in methods to cheat. They were finding tools like QuillBot to disguise AI-generated writing and make it sound more like student writing.

She’s open to going back to the classroom, if she gets more administrative support to help detect the kind of AI misconduct that felt out of control when she left.

“As a community, we need a support network,” she said.

Lily Altavena is a national reporter at Chalkbeat. Contact Lily at laltavena@chalkbeat.org. Lily is on Instagram, Tik Tok, and Facebook.

Wendell Berry, who died at his home in Kentucky yesterday, was a rare combination of poet and farmer. That wasn’t always an unusual thing in the literary world, but the authenticity of agricultural and pastoral poetry has been in decline since the days of Hesiod. Even Virgil was a bit of a poser. But Berry was the real deal in every way.

That would set him apart in any age, but especially in our own time, when so many of us (writers included) are cut off from nature and the holistic ecosystems of agrarian life.

To mark his passing, I’m sharing this article from my archive on Berry’s nine rules for technology. It’s a simple list, but a profound one—and even more relevant today than when it was first published in 1987.

If you want to support my work, the best way is by taking out a paid subscription (just $6 per month).

Subscribe now

Back then, Wendell Berry was living on a farm in Kentucky. He did his writing with pen and paper, and his wife Tanya would create typewritten drafts of his manuscripts on a Royal standard typewriter purchased in 1956. That manual tool was, he insisted, “as good now as it was then.”

I’m sure he said the same thing about the implements he used to sow and reap. But friends told Berry he needed a computer. It would make it easier to write, they insisted.

In response, Berry came up with his list of nine reasons to embrace new technology. Let’s revisit them, one by one.

Nine Standards for Technological Innovation

(1) The new tool should be cheaper than the one it replaces.

This is a very persuasive selling point for new technology. And for most of my life, tech companies worked hard to lower prices.

I still recall my parents scrimping and saving in order to buy a color television when I was seven years old. It cost almost $500—a huge amount in those days.

They probably should have waited. A few months later, RCA dropped prices to $399. Prices continued to drop in later years. You can buy a high tech TV today at Best Buy for less than what my parents paid in the 1960s.

Computers also got more affordable—at least until recently. I got my first computer (an Apple IIE) when I was in graduate school—it was an expensive gift from the Boston Consulting Group in exchange for accepting their job offer.

The list price back then was $1,400. I could never have afforded to buy it on my tight student budget.

But, over a period of many years, each subsequent computer I acquired was better and cheaper than my previous model. Alas, that happy trend has now ended.

When I buy a new computer now, I pay more. And the performance is not always better. I recently had to scrap a new desktop after only a few months, and go back to my previous model. The new computer didn’t work as well as my five-year-old one.

When did new tech stop getting cheaper and better?

I have a cranky answer, but an accurate one. It happened the day Steve Jobs died. Maybe not exactly on that date—but shortly afterwards.

Look at this chart of iPhone prices, adjusted for inflation, and you can see what I mean.

Now let’s go to the second reason to adopt new tech from Wendell Berry’s list.

(2) It should be at least as small in scale as the one it replaces.

This is another good reason to upgrade your setup. And tech did get smaller for many decades.

Guess who played a key role in that? Yes, Steve Jobs again. Because of his obsession with product design, we now carry a huge amount of advanced tech in our pocket.

Just consider this remarkable fact: Every device featured in this Radio Shack advertisement from 1991 has been replaced by your tiny phone.

But this, too, changed soon after Jobs died. (Are you noticing a pattern here?)

The thinnest iPhone ever was the iPhone 6 (2014)—at a slim 6.9mm. The company continued to launch ‘mini’ models for a few years, but stopped after iPhone 13.

Tech is now bulking up. It’s not just the devices—wait until you see those AI data centers. A single facility can spread over two kilometers.

(3) It should do work that is clearly and demonstrably better than the one it replaces.

This is the most obvious requirement for new tech. It needs to work better than old tech.

But Silicon Valley has totally abandoned this ideal. Every web interface I use has gotten worse over time—from search engines to social media to software to shopping apps.

Google is worse than ever. Social media is worse than ever. Amazon is worse than ever. Facebook is worse than ever. Everything I get from Microsoft is worse than ever.

So here, too, we see that new tech previously fulfilled Berry’s requirement—but stopped doing so around the time Steve Jobs died.

(4) It should use less energy than the one it replaces.

Here, again, we see an ominous reversal. With the rise of AI, tech companies now use up more energy than ever before. They are sucking the power grid dry in many places.

And it’s going to get worse—much worse.

What makes this especially revealing is the fact the public intensely dislikes AI—surveys make this absolutely clear. So tech companies are destroying the environment solely to increase their dominance and control—not to please you and me.

(5) If possible, it should use some form of solar energy, such as that of the body.

Now Berry is asking for something our technocracy has never delivered. And here we encounter the exact opposite of the AI situation described above. The hot new plan is to construct nuclear reactors on site. What could possibly go wrong?

There’s another striking contrast here. AI depends on huge investment from corporations, while consumers are mostly indifferent. Solar energy is the opposite: It’s supported by investment from consumers—who use it to heat their homes, water, etc.—while corporations are mostly indifferent.

What a sad state of affairs. Private citizens have more prudent approaches to tech than the tech companies themselves.

(6) It should be repairable by a person of ordinary intelligence, provided that he or she has the necessary tools.

This, too, has changed during my lifetime. I once saw my father unscrew the back of our home TV set, and fix a malfunctioning part. Nowadays you can’t even open up those bad boys.

Tech providers create all sorts of obstacles to prevent repairs—unusual screws, arcane software, special tools, etc.

Consider the case of John Deere tractors, which wouldn’t start until a company-trained technician cleared out the error code. The company also refused to sell spare parts. Their practices got so abusive that politicians passed right-to-repair bills to protect farmers.

But the worst example happened during the COVID pandemic, when companies tried to prevent hospitals from fixing their malfunctioning ventilators. Manufacturers put software locks on this life-saving equipment to prevent repairs.

This represents a total failure on the part of the technocracy—and actual malfeasance by the executives who run these companies.

(7) It should be purchasable and repairable as near to home as possible.

Finally I can give some tiny credit to our tech titans. They do offer home delivery—even if the product is made in a sweatshop far, far away.

(8) It should come from a small, privately owned shop or store that will take it back for maintenance and repair.

This is a pipe dream. The tech product lifecycle is built on planned obsolescence, not simple repairs.

When your device or software stops working, you replace or upgrade—whether you want to or not.

In some instances, you aren’t even allowed to own, let alone fix, your tech—you just license or lease or subscribe. It’s almost like capitalism in Silicon Valley has turned into communism. You will own nothing, and will love it.

(9) It should not replace or disrupt anything good that already exists, and this includes family and community relationships.

This may be the biggest tech failure of them all.

The leading tech companies have deliberately promoted dysfunctional apps that destroy lives. And they know it.

• Leaked internal documents from TikTok show that they were aware that teens get addicted to their app in just 35 minutes. They built it that way.

• Facebook knew that Instagram use leads to depression, anxiety, eating disorders, and other problems.

• Spotify insiders have confirmed the company’s systematic plan to reduce royalties to musicians by manipulating passive listeners.

• For more examples, see my list of 52 indicators that technological progress is reversing.

This is the new normal for tech: It deliberately makes things worse, not better.

Here’s the entire list of Wendell Berry’s criteria. If this were a report card, your tech leaders would all get failing grades.

The curious fact is that the most up-to-date and forward-looking thing is this whole article is Berry’s list from 1987. Nothing on it is obsolescent or inappropriate or dysfunctional or harmful.

I wish our tech companies could say the same for their work. Maybe they could learn a thing or two from our poet-farmer—or at least take a look at his list. We can only hope.

Universities across the country are investing in artificial intelligence literacy. Students are attending workshops, experimenting with generative AI tools and earning certificates meant to show that they are ready for an economy shaped by AI.

The problem is that we don’t know if these programs help students solve workplace problems, if they open doors to job opportunities, or both. We also don’t know if they give students from different institutions a fair chance to compete.

Higher education has an AI pilot program problem. Much of the evidence for “successful” programs still comes from single-region studies and relies on short-term measures such as test scores and self-reported data. Too often, they are described as successful before anyone can show whether students have actually used what they learned beyond the classroom.

Colleges should stop treating AI literacy as the finish line and start building clear pathways from learning to workplace application. Every AI literacy program should be designed with employers, tested across institutions and evaluated by what students can actually do afterward. Attendance, certificates and satisfaction are not enough. The goal must be readiness.

Related: Interested in innovations in higher education? Subscribe to our free biweekly higher education newsletter.

AI literacy is necessary. Still, just knowing how to use a tool does not mean a student knows how to apply it responsibly to an unfamiliar workplace problem. A student may know how to ask a chatbot a question without knowing how to check the answer, protect private information, explain a recommendation or decide when the tool should not be used. Those abilities develop through repeated practice in real contexts. That is why colleges and universities should adopt shared measures of AI workforce readiness.

The stakes are especially high for students at public and less-resourced institutions. They may be introduced to AI tools without receiving enough practice, guidance or employer connections to turn that introduction into an opportunity.

Earlier this year, I founded an initiative to examine this challenge across two different settings. The program brought CUNY undergraduates and NYU Tandon graduate engineering students into the same hands-on series of workshops. Industry professionals showed students how AI is used in actual jobs, and students practiced using the tools for workplace tasks and explaining their decisions.

Our small, preliminary study found that the two groups reached similar levels of applied AI use. Participation emerged as the clearest factor in predicting that use. Among students who attended one or two workshops, regardless of their school, 21 percent created something with AI. Among those who attended six or more, 56 percent did. The study needs to be repeated elsewhere, but it did not detect a difference between the two institutional groups on this measure. The design of the program, along with the amount of meaningful practice students receive, may matter even more.

Related: OPINION: Schools cannot teach AI literacy without a way to measure it

A useful question is whether particular combinations of instruction, practice, professional guidance and real-world opportunity produce meaningful results, and which students benefit. Answering that question will require universities to move beyond isolated pilots.

First, to evaluate their AI literacy programs, colleges should measure changes in students’ judgment, problem-solving, professional communication and ability to create useful work with AI. They should also track whether students gain access to internships or employment.

Second, universities should study programs across different types of institutions. Findings from any one program cannot automatically be assumed to generalize across different institutional contexts.

Third, employers should become partners in pilot program design rather than occasional guest speakers. They can help define real problems, review student work and explain which abilities matter on the job. Students need contact with professionals who can show them what responsible AI use looks like inside an actual organization.

Encouragingly, federal workforce policy is beginning to move in this direction. In July, the U.S. Department of Labor awarded nearly $162 million through five agreements to expand Registered Apprenticeship programs. The funding uses performance-based incentives tied to results such as the hiring of new apprentices, keeping them in programs and helping them advance.

Jobs for the Future, for example, a national nonprofit, received $40 million to support Registered Apprenticeship growth in roles building and maintaining the critical infrastructure that sustains the artificial intelligence, semiconductor and nuclear energy industries.

But apprenticeships are not university courses: They pay wages, provide federally recognized credentials and place learning inside a job. While colleges cannot simply copy that model, they can work toward the same goal: training that leads to demonstrated skills, meaningful employer participation and measurable outcomes.

Since generative AI reached college campuses, universities have shown that students are interested in learning how to use it. The next phase should determine which programs actually help them use AI responsibly and carry those skills into the workplace. Future grants should require shared measures, cross-campus comparisons and follow-up after students leave, since no single campus can build this evidence alone.

Without that evidence, AI workforce readiness will remain a collection of promising campus stories. With it, higher education can build a system that works for students no matter where they enroll.

Ngoc Cindy Pham is an associate professor of marketing at Brooklyn College, CUNY, and a visiting research professor at NYU Tandon. She is also a Fulbright Specialist and founder of BRIDGE AI Lab.

Contact the opinion editor at opinion@hechingerreport.org.

This story about college and AI was produced by The Hechinger Report, a nonprofit, independent news organization focused on inequality and innovation in education. Sign up for Hechinger’s weekly newsletter.

The post OPINION: Knowing how to use AI tools is not the same as knowing how to apply them responsibly. Colleges must do more appeared first on The Hechinger Report.

Worms, Alter Ego, MacJesus ProGold, Portal (1986), and many more

Most educators agree that there should be some guardrails regarding how students use AI in their independent work. Here, I argue for this guiding principle: Students should only use AI for things that they already know how to do well.

A story is told about Harold Alexander, British field marshal in command of the evacuation at Dunkirk. At the end of the workday he had the habit of taking any correspondence still in his “in” tray and simply putting it in the “out” tray. When his assistant once asked about it, he said “You’d be surprised how little of it comes back.”

That sums up my approach to the winds of educational change. When a new technology appears and people are asking themselves “what about my teaching should change?” my instinct for my own teaching is “change nothing,” partly because I suspect that the excitement is overblown, and partly because I’m happy to let more venturesome people make a change first, and learn from their experiences.

But that strategy can’t be applied in the case of AI. In 2023 I, like every other educator, realized that I had to respond to student access to large language models.

Administrators have had a few years to think through how teachers should change classroom practice to meet the new reality, and many districts have posted guidelines. The policies I have seen have much in common, and I think they are wrongheaded.

(Throughout this post I’m referring to students’ independent use of AI, not to educational software nor to teachers using AI in their work.)

There seems to be widespread agreement that there’s a problem, and even on how to characterize it. Learning is a product of engaging in mental work, and students may use AI to replace that mental work and hence, not to learn. The question is what to do about it.

Almost no one advocates for “no restrictions on independent AI use” nor for “No independent AI use by students. Period.” The overwhelming majority have taken the position that students should use AI, but there should be limits.

Mashable recently ran an article summarizing the response of the US’s largest school districts. Here are the one-sentence versions, as I read them:

New York City: students are allowed to use AI for basic “research, exploration, and creative projects.” There must be educator oversight.
Los Angeles: No AI use under age 13. Over 13, students can use AI to brainstorm and to edit their text, but students must produce text. Students must cite what AI contributed.
Chicago: Students can use AI for tasks like brainstorming and summarizing. They are encouraged to use AI as a study partner.

The policy of my home county (Albemarle, Virginia) is similar. There are published categories of AI support from which teachers may choose.

In short, teachers are given a good amount of leeway to allow students to use AI or to forbid it. There is, however, little or no guidance about how that choice should best be made.

These policies share a vibe. The goal is that students use AI as an associate in their work, but the student is still very much the lead. The second goal is that students be honest about exactly what AI has contributed.

This lead/associate orientation makes sense for life after graduation. Once a student is in the workforce, they will use AI to produce products: new ideas, reports, and so on. A guiding principle of “Use AI, but let me see your work” makes sense.

This is not, however, an optimal guiding principle for students. Students produce products, but the product is seldom the point. No one wants to read the papers they write, and the math problems they solve have no practical value. Teachers have them do these tasks because they provide practice in cognitive abilities that we think matter.

Writing is an especially interesting example. Writing a substantial paper requires integrating what you know across multiple sources, formulating a detailed argument, articulating your thoughts precisely, anticipating what your reader knows and can understand, and more.

For many educators, writing papers was the only tool we had to ensure that students engage these cognitive processes. AI took that tool from us, and we don’t know of a good replacement.

This loss was unprecedented. The closest analog might be math problems and calculators or Google Translate for foreign language. But in those cases a teacher could still remind students that they would not have these tools available during in-class assessments. I can’t give my students an in-class assessment that is cognitively comparable to writing a ten-page paper.

So the point of assignments is the mental processes required to complete them, and the point of the mental processes is learning. That seems to suggest a simple litmus test for the use of AI. Artificial Intelligence tools should not substitute for tasks wherein students would benefit from doing the mental work themselves.

So…do students learn when they brainstorm? Do they learn when they edit their work? Of course they do. This principle indicates that students should use AI only when doing the work themselves would yield little learning. This guideline echoes how most of us think about calculator use. Once you’ve mastered arithmetic operations, there’s no benefit to hand calculations when you’re solving an algebra problem. But if you’re still learning algebra, don’t use the algebraic functions. The slogan might be:

“Only use AI for what you already know how to do.”

I can see two objections.

First, can’t AI scaffold learning? Part of the learning process is doing things badly. Can’t AI offer pointers for improvement?

I see two problems with this idea. The first is that the novice is not going to be a very good partner to AI in learning. The novice doesn’t know enough to ask good questions. He will give the AI system a vague prompt about the goal of the exercise whereupon AI will sharpen it for the user. Essentially, it will chivvy the user toward the polished product, and the novice will simply accept the suggestions. (Remember, I’m talking about independent use of LLMs, not about a product custom-made to provide instruction. That’s a different matter.)

The second problem I see with AI-as-scaffold is motivational. Sure, students could learn from AI by asking it to critique their work, to generate counterarguments, and so on. And I’m sure some would, some of the time. But I think it’s asking a lot of students—or better, asking a lot of human nature—to expect them to take a much harder mental path for a project in a typical curriculum, even if it is in the interest of their own learning.

A second objection to the principle might be this: even after you reach competence, you still need practice; that will deepen and refine the skill and/or knowledge. For example, if a ninth-grader can write a good paragraph, is it now okay for AI to write paragraphs for the student?

If students are competent, a teacher should ask whether doing the task themselves is still part of what students are meant to learn or practice. When students are writing, editing, reasoning, problem solving, or brainstorming (for example), the mental process is usually the thing. You don’t want the help of AI because it replaces the mental work that contributes to learning.

The alternative is that AI is replacing a task that merely supports the actual mental work you want the student to do; the replaced task was soaking up the students time and energy and providing no benefit. That would be the college student hand-calculating long multiplication problems required for her engineering problem set.

For this reason I add the word “well” to the candidate principle: “Students should only use AI for things that they already know how to do well.”

Now of course there’s a judgment to be made here. A physics teacher might think “It’s great that AI will polish my students’ writing for their lab reports“ whereas someone else might argue “your students need to learn to write in your class too.” But of course this problem is not new. We’re just seeing the issue in a new guise.

We’ve all heard “They’re going to have to learn to use AI.” But I think that slogan applies to how students will use AI at home and in the workforce, where the product is the thing. For students, learning is the thing. That’s how I come to the principle, “Students should only use AI for things that they already know how to do well.”

"A register of the products that became the default in their trade." Early days. Largely US-focused.

Permalink