Tomorrow I hand a room of Georgetown graduate students a syllabus I spent the summer tearing down to the studs and building back. Why? About half of what I used to assign could be finished in seconds by a chatbot.
I have kids in elementary school, in high school, and in college. So I've read this year's education polling twice over: once as the parent, and once as the professor.
Fifty-two US surveys on AI in the classroom came through Poll Vault in the past twelve months. They point to three groups - teachers, students and parents - making it up as they go along, and each convinced the other two are the ones getting it wrong.
Parents Just Don't Understand
Lake Research and Echelon put the same question to about 840 parents and 800 of their teenagers. A majority of the parents, 52%, called using AI on schoolwork unethical, the sort of thing that should carry a consequence. A majority of the teenagers, also 52%, called it innovative, the sort of thing school should reward. Imagine what those dinner conversations are like!
Ask teenagers what they mainly use AI for and 59% say looking things up; ask their parents to guess, and only 42% land on it. Parents overshoot in the other direction on loneliness: 23% assume their kid turns to chatbots for company, against the 8% of teens who say they actually do. DJ Jazzy Jeff and the Fresh Prince had it right: parents just don't understand.
The Kids Are Ahead
Whatever the grown-ups settle on, the kids are already a step ahead.
86% of children between nine and seventeen have used generative AI, Echelon Insights found in a May survey of 1,200 of them. A quarter use it every day, and among those daily users, 42% say going a month without it would be hard.
The older they get, the more they run on it. Pew finds 54% of teens using chatbots for schoolwork, and one in ten who say they do all or most of it that way. 59% say students at their school use AI to cheat at least somewhat often, and about a third say it happens all the time.
And AI is disintermediating the child-adult educational relationship. In Common Sense Media's survey of nine-to-seventeen-year-olds, nearly one in four said that when they're stuck on homework, they'd go to a chatbot before a teacher, a counselor, or a parent.
But what makes this so interesting is that despite these high levels of use, young people are more skeptic than enthuasiast. California State University asked 94,000 of its students and faculty about AI, and while 64% say it improved their own learning, 65% say AI has not improved education. In their minds the tool helps them and hurts the institution.
Similarly, the Digital Education Council's study of 45,000 students across 35 countries finds two-thirds afraid AI is making learning too shallow, a share that climbs to 81% in the US and Canada. 43% of North American students say they'd support banning it on campus outright.
This could be driven by looming anxiety about the job market they are about to enter. 65% of Harvard undergraduates told the Crimson (a survey of 303 of them) that they expect AI to hurt their job prospects. Lumina and Gallup found 16% of college students have already switched majors because of AI, a number that climbs past a quarter in vocational and tech fields. They are not wrong to worry. 65% of technology hiring managers told ResumeTemplates they'd sooner put money into AI than into a 2026 graduate.
An Asset for Me, a Risk for Them
The teachers have their own perspective.
Georgia's Department of Audits surveyed 13,679 public-school teachers. Nearly 60% now hand some of their own work to AI, mostly lesson plans and parent emails, and 89% of the ones who do say it works well. Ask those same teachers about their students using it and the answers change: plagiarism, lost collaboration, thinking that never gets exercised. It's a productivity tool in teachers' hands and a threat in students'.
Colorado's Department of Education caught the split in one pair of numbers from 42,000 educators: 66% use AI to plan their teaching, and 32% feel prepared for what it's doing to their profession. Only 34% have any method for telling whether the essay in front of them was written by a person.
That last number is the real bind. You have thirty papers to grade and no detection tool you trust. Three of them read cleaner than the drafts you saw in class — arguments a little tighter, the vocabulary shifting somewhere around the fourth paragraph. Your school has handed you no policy and no training. The student swears the work is theirs. What do you do? (Asking for a friend!)
The same Instructure survey that found instructors untrained also found their students all the way in: 90% of college students use AI in class, against 61% of the instructors. The people writing the assignments are twenty-nine points behind the people turning them in.
The Only Rule is There Are No Rules
Everyone's improvising, and it's because so few school administrations have taken the time to write the rules.
Only 33% of parents say their child's school has any AI guidelines, and 38% don't even know whether one exists, per Deloitte, in a survey where 49% say they worry their kid already leans on AI too much.
The National Parents Union, surveying 1,511 K-12 parents, found 47% had received no information at all about their school's approach, and 57% had never once been asked what they thought. Higher ed is no better organized. Just 10% of college students rate their institution's response as handled "very well," another 27% as "fairly well," and the largest single group, 34%, call it inconsistent, according to Inside Higher Ed.
Kids adopted before teachers. Teachers adopted before schools. Schools adopted before anyone wrote a rule.
This year Poll Vault tracked 52 US surveys on AI in education — see them all →
Where the Line Actually Is
I should say where I stand. I'm an optimist about this technology, and nothing in these fifty-two surveys reads to me as a case for keeping it out of school. It reads as a case that the tool showed up before the grown-ups did their jobs. We can fix this because we've been here before! Every classroom technology back to the calculator arrived ahead of its rules.
But rules work! Among UK students who use AI, 67% told Edinburgh Napier they wouldn't touch it on an assessment they'd been told to keep clean, and 32% don't use it for coursework at all.
And voters are clear on what they want out of those rules: Change Research talked to 2,006 voters, and 85% said students should not use AI to write their assignments or generate their answers — Democrats and Republicans within five points of each other. But solid majorities want AI in the room as help: 63% to explain a hard concept, 56% to catch grammar, 52% to study for a test. Help them think, yes. Do the thinking for them, no.
This is where I've taken my own course. It's no longer about mastery of coding syntax - a chatbot can easily generate it for you. Instead its about judgment, taste, perspective and being able to defend the choices you made in front of the people that will push you on it.
That means a course heavier on theory so that it can be used as a directive to AI agents. It means using the LLM to pressure test your assumptions or instructing a chatbot how to have it review your work. It's no longer about mastery of the machine through syntactic fluency; now it's about your management of and collaboration with your AI co-worker.
Every school in the country is going to have to draw some sort of line, and most of them haven't started. The data says that when they finally do, the public will be behind them — and so will the students, who have been asking for some clarity on this for a while.
As for whether I'm providing any clarity with my class? We'll find out tomorrow!
Quick Hits
📵 The one thing everyone agrees on. 80% of voters back banning or restricting phones in schools: 84% of Democrats, 91% of Republicans, and 68% even among 18-to-34-year-olds. Change Research
🎓 The presidents know, about everyone but themselves. 52% of college presidents say AI literacy isn't widespread on their campus, and the same share say the sector is unprepared. Only 29% say that about their own institution. Hanover Research, n=430
👩👧 Mothers are near-unanimous. Just 6% of mothers see AI as a positive force kids should use freely, and the unease runs across conservative, moderate, and liberal mothers alike. 43% say their child already uses it for school. Count on Mothers, n=2,290
🏫 The faculty are voting with the syllabus. 52% of CSU faculty say AI has hurt their teaching, and 40% now discourage or outright ban it in their classrooms. California State University, n=94,000
📉 The public verdict. 46% of US adults say AI will make children's education worse; 23% say better. At the intense end it's 30% to 8%. YouGov, n=9,242
What Else We Tracked
- Three-Quarters of Parents Back AI in Schools, But Tech Dependence Is Top Fear — SurveyMonkey
- Half of Parents Fear Kids Rely on AI Too Much as School Use Outpaces Guidelines — Deloitte
- Massachusetts Parents Split Three Ways on AI in Schools — MassINC Polling Group / EdTrust
- Majority of US Teens Say Peers Use AI to Cheat; 1 in 8 Turn to It for Emotional Support — Pew Research Center
- Two-Thirds of Teens Fear Deepfake Exploitation as AI Explicit Content Spreads — Common Sense Media / Burson
- Half of Campus Tech Leaders Say AI's Return on Investment Is Unclear or Disappointing — Inside Higher Ed / Hanover
- Nearly Half of High Schoolers Use AI in College Search, But Distrust AI-Generated Outreach — EAB
- Students Confident With ChatGPT but Unprepared for Industry-Specific AI Jobs — Autodesk
Until next time,
Alex Lundry
Due Data is powered by Poll Vault. Get the full data behind every poll mentioned above.
Want the full data behind every poll mentioned above?
Create free account →