AI helps kids do better in school. That doesn’t mean they’re learning

The best way for us grownups to prepare for AI is to use AI, I’ve frequently argued. But when it comes to the benefits of kids using AI, there are far more questions than answers.

New York City’s just-announced AI “moratorium” for elementary and middle-school students is among the most striking manifestations of the growing worries over AI’s growing presence in classrooms. Yet there’s also concern it may be too little, too late.

According to a 2026 Pew Research survey, 54 percent of students reported using AI for help with their schoolwork, with 12 percent saying they got AI help with all their assignments. A 2025 report from the Center for Democracy and Technology goes even further, estimating that 85 percent of teachers and 86 percent of students used AI in the 2024-25 school year.

OK, so what’s the lowdown about students and AI: Does it help them perform better in school, or does it harm their ability to learn? Actually, it does both, according to the latest research.

Welcome to another edition of Prompt Mode, your weekly AI newsletter.

I’m your host, Ben Patterson. Each week on Prompt Mode, I’ll be serving up analysis of the AI trends that matter to everyday users like you and me. Stay tuned for practical AI tips, hands-on experiences with the latest AI tools, and–you guessed it–prompts to help you get the most out of your AI assistants.

Thanks for reading, and if you like what you see, just sign up right here.

A recent Stanford University study showed that K-12 students frequently perform better on math, writing, and coding tasks when using AI tools. But when you take the chatbots away, those improvements tend to evaporate, leaving some students (especially those using straight-up AI chatbots rather than those fine-tuned for learning) more dependent on AI tools than truly helped by them.

Another study from the Brookings Institute found that while “well-designed” AI tools used in an “overall, pedagogically sound approach” can be a boon in the classroom, an “overreliance” on AI chatbots can harm a student’s ability to learn and hinder their relationships with their teachers and fellow classmates.

The AI ban for New York City schools, which calls for a one-year pause in “student-facing generative AI” for public school students through the eighth grade, is among the toughest in the nation, yet some fear it’s still not strong enough. For one thing, the current ban only encompasses a single school year, not two as many parents had hoped, the New York Times reports.

There are also some key loopholes, including a provision that “centrally approved” e-books and coding programs are exempt from the AI ban, the Times notes.

Of course, there have always been debates about computers in the classroom. Personally, I love that my daughter (who’s starting her first year of public high school in Manhattan, where AI in the classroom is still allowed) has been studying Scratch coding in class. But while I’m pleased she’s working with computers, I mainly want her to learn how to learn, rather than for her to learn about learning with AI.

Even if it isn’t perfect, I think New York’s pushback on AI in the classroom is a step in the right direction, as well as a discussion worth having. But like all things AI, I wonder if we haven’t already passed the point of no return.

Prompt of the week: The “verify the name” prompt

This week’s prompt is a bit different from the others. Written by Anthropic specifically for Claude Fable 5.1, the latest and most powerful (for now) Claude model, this prompt tackles a specific Fable 5.1 quirk: its tendency to skip searches and rely on its memory during low-effort sessions.

The Anthropic prompt “nudges” the model to double-check its facts using its search tool rather than making assumptions based on its training data — and indeed, it’s a prompt that would do plenty of other AI models good, especially those that run locally.