If you walk into almost any high school in the United States right now, you’ll notice that something has changed. It happened quietly, without a report or policy memo. When students sit down with their laptops and type a few words into a browser tab that isn’t the school’s learning portal, a full paragraph shows up right away. There was no rule that said this was okay. Also, no one really said it wasn’t. This is where American education is right now: that gap, that long, cold silence.
Based on surveys done between June 2024 and June 2025, new research from the College Board backs up what many teachers have already thought for a while. The number of high school students who use generative AI tools for schoolwork went up from 79% to 84% in just four months, from January to May 2025. That’s no longer a weird trend. That’s most of a generation.
It all revolves around ChatGPT. As of May 2025, almost seven out of ten high school students said they used it for homework and assignments. They use it to come up with ideas, edit papers, and find sources. Some people are writing things on it that they used to write themselves. Sometimes it’s hard to tell the difference between help and replacement, and a lot of students might not be looking for it.
Why is this strange? Because adults in the system can’t agree on how to feel about any of it. Data from the College Board shows that more than 85% of school administrators think it is really helpful for students to learn how to use AI. At the same time, those same administrators—principals, AP coordinators, and teachers—are becoming very worried about academic honesty and whether or not students are really learning the basic skills they’ll need later on. Most schools are stuck in a contradiction they can’t get out of.
To be honest, policy is a mess. A little over half of principals say their schools have not blocked any generative AI tools on school networks. About two out of five schools or districts don’t let students use it, but it’s clear that they do anyway, from their own devices, on their own time, and often in ways that no one is watching. Only about 13% of schools actively encourage all of their classes to use AI. Another 16% leave it up to each teacher individually. As a result, two students in the same building who are sitting in classes next to each other are following totally different rules.
Talking to people who work in education makes me think that the students have already decided whether something should be banned or allowed. There was a strong message sent when New York City and Los Angeles school districts blocked ChatGPT from their networks in 2023. But when they got home, the students opened their own laptops and continued to use them. You can’t stop an idea or a habit just because you block a website on school Wi-Fi. It just takes it off campus and out of sight.
If you pay close attention, something more subtle is what worries experienced teachers more than cheating. One half of high school students aren’t sure if the pros of AI outweigh the cons or if they don’t. They’re stuck in a state of ambivalence, which might be the most honest answer. They are working with a tool that they don’t fully understand for an organization that doesn’t fully understand it either. One day, people will expect them to know the difference between thinking and outsourcing.
Still, it’s not clear if this time makes a generation smarter or just more capable by giving them better tools to fill in the gaps. They’re not the same. As American schools still deal with tight budgets, a lack of teachers, and gaps in their facilities, they are being asked to answer this question right now, without a guide.
The classroom after ChatGPT is definitely not coming. It’s already here, and it’s running on questions, trials, and a lot of emails that haven’t been answered between administrators and curriculum teams. That was something the kids knew before the adults did. Often, they do.
