As artificial intelligence begins to reshape the economy, the next question is unavoidable: how, if at all, should schools bring it into the classroom? 

America’s largest school districts have taken dramatically different approaches. New York City has barred student-facing generative AI through eighth grade, while allowing limited high-school pilot programs. Los Angeles has imposed a districtwide moratorium on student generative-AI use on school devices. Meanwhile, Alabama has moved in the opposite direction, incorporating computer science, internet safety, and foundational technology skills into instruction beginning as early as kindergarten.

So what is the answer? 

In an interview with Fox & Friends First, Manhattan Institute K–12 Education Policy Fellow Jennifer Weber offered a more sensible answer. She rejected outright bans on AI, favoring an approach closer to Alabama’s, but argued that AI instruction should not begin until third grade, after children have first had time to develop the basics of reading and writing.

"If we ban it, we're not banning it from their entire environment. We're banning it from schools. And schools are where we're supposed to be teaching kids," Weber said. "So if we just do a ban for the school day, we're just giving them access to it outside of the school day. And the kids that are not in public schools are probably going to learn to use it. And so the kids that are in public schools are going to be the ones that are left behind."

Recommended
It Turns Out California Runs a Shadow Welfare System for Illegal Aliens Amy Curtis Beat the Press Kurt Schlichter

"How did Alabama really thread the needle on this?"

"They had a plan. They are starting early, but they're not giving a chatbot to kids in kindergarten either. I think that's what's very misunderstanding of it. They're teaching kids what the Internet is, understanding public versus private information. They're teaching kids the foundation of technology," she said. "And then later on, they'll give kids AI. But first, they're integrating it into their K to 12 system. So it's part of it, as opposed to just saying, we don't want it and we're going to ban it."

"But you would agree that kindergarten is a little young to start this, correct?"

"Oh, absolutely. I think that K to 2, we need to focus on teaching kids to read," Weber replied. "There are way too many kids in the United States that can't read right now. And you need to read to be able to eventually use AI well. You also need to be able to ask questions. You need to be able to determine fact from opinion. So there's a lot of teaching that needs to happen. So I don't think giving kids a chatbot in kindergarten is definitely not the solution. We should put pressure on teaching those foundational skills in K to 2. And then starting in grade 3, start to slowly integrate it into it, because that is the future."

The concern over bringing AI into public education is legitimate. Research involving more than 26,000 students found that AI can raise homework scores and reduce the time students spend completing assignments, while their later exam performance falls sharply. That is less an indictment of AI, and more an indictment of how students are prone to use it.

However that does not make AI inherently incompatible with education. Used under clear limits, it can help students practice, research, receive feedback, and learn to evaluate the strengths and failures of a tool they will use throughout their entire adult life. The alternative cannot be a choice between unrestricted access and total prohibition. A policy of letting AI replace student thought is irresponsible; a policy of pretending it can be kept out of students’ lives is simply not serious. 

The task for schools is to teach children to use the technology, not to let the technology use them.