Panic drove NYC and LA to ban school AI — and that’s a big mistake

There are two ways for America’s schools to get artificial intelligence wrong.

One is to hand a child a chatbot and let it do all the thinking — and the other is to ban AI and pretend it doesn’t exist.

New York and Los Angeles have both chosen the second option.

That means they’re setting their students up to get the first.

Americans are split when it comes to AI, as evidenced by the debate that raged last week.

When the heads of Anthropic and OpenAI urged AI companies to slow development down, President Donald Trump dubbed their alarm bell a “sick conspiracy” — and skeptics saw self-interest behind both warnings.

Mayor Zohran Mamdani faulted Trump for what he called a “very flip response” to AI companies’ talk of “the end of humanity,” and the New York City Council has called for an all-member hearing to question AI companies about the dangers.

The nation is somewhere in between: A Politico poll this month found that 48% want to pause the development of advanced AI. 

Yet on Sept. 2, Mamdani and Chancellor Kamar Samuels chose to go to extremes, banning generative AI in NYC public schools for every student from 2-K through grade 8 — about 600,000 kids.

Later that same day, Los Angeles blocked AI on school devices in every grade — banning it indefinitely for more than 378,000 students.

These school systems are behaving as if a kindergartener’s educational needs are the same as a 13-year-old’s, or a 17-year-old’s.

But their bans fail no matter who is right about AI.

If Trump is right and AI is the engine of the future economy, our students must learn how to use it.

If Mamdani is right and AI is dangerous, students must learn how it works, so they can understand its flaws and pitfalls.

Either way the answer is to teach — and a ban teaches nothing.

Mamdani calls his policy a moratorium. But a moratorium only works if the world outside is also pausing, and right now it isn’t.

More than half of US teens already use chatbots for schoolwork, according to Pew, and one in 10 students say a chatbot does most of it.  

So students with AI access at home will learn through trial and error, not through careful instruction.

And the rules don’t bind private schools, charter schools or tutors, nor do they touch what happens at home or across the rest of the country.

That means students without access will fall behind.

Consequently, New York’s and LA’s moratoriums aren’t taking AI away from kids — just stripping them of the opportunity to learn from teachers how to ask AI good questions, evaluate its information, and gauge the trustworthiness of its assertions.

This isn’t to say students at every level should have access to AI in school.

For the youngest students, through second grade, the ban is correct: They must learn to read first.

However, beginning in third grade students should be taught to read what AI produces, spot weak claims, and distinguish fact from opinion.

Too many students today can’t do that.

When Stanford researchers tested 3,446 high school students in 2019, two-thirds of them could not tell a news story from an ad, even when the ad was labeled.  

Early research shows that this kind of work can improve critical-thinking skills.

By middle school, a student needs to be able to trace an AI’s claim back to a real source and spot a citation that doesn’t exist — a skill that takes years to teach. 

Yet the nation’s two largest school districts have no plan to change their teaching to include what students should learn about AI at different grade levels, or when.

That’s outright negligence.

As I’ve argued in a Manhattan Institute brief, students need a real AI-literacy curriculum, one that teaches them to judge what AI produces so it never does the thinking for them.

Regarding AI as a tool, rather than a threat to learning, is key: More than a third of entry-level jobs already require AI literacy.

That’s also what a cautious public is asking for.

Americans are not telling schools to hand children over to a chatbot — but neither are they telling schools to pretend AI doesn’t exist.

A curriculum that keeps AI out of the earliest grades, puts the teacher in charge in third grade and builds real skills by middle school respects that caution and still prepares students to succeed.

Students are already using AI, whether the two largest school districts like it or not. 

We can teach them how to use it well — or leave a million kids to just figure it out on their own. 

Jennifer Weber is a fellow at the Manhattan Institute.