The future of AI in the classroom

I spent most of the last two weeks on the road in California at three education conferences that highlighted the risks – and potential benefits – of artificial intelligence in the classroom.

The American Educational Research Association (AERA) and the National Council on Measurement in Education (NCME) events were held in Los Angeles. After they wrapped, I headed for San Diego and ASU+GSV where I moderated a panel on artificial intelligence tutors.

The problem with humans in the loop

While gathering ideas for future columns, I heard two things at the NCME convention that stuck with me.

• We all know we're supposed to check what AI spits out. But Victoria Yaneva, director of data science and AI at the National Board of Medical Examiners, said there's growing evidence that humans are getting worse at doing so. People who are enthusiastic about AI are more likely to miss errors, she said. AI skeptics are better at catching them.

The people most wary of AI may be the ones we most need in the loop.

• I heard a review of roughly 250 studies on AI-generated test questions and scoring. Only one - one study - looked at whether the results were biased against certain students.

John Whitmer of Learning Data Insights called that oversight by the research community a "big miss."

Related: Asian American students lose more points in an AI essay grading study

Ed tech's AI dreams hit reality

The ASU+GSV Summit in San Diego is a giant ed tech gathering where investors, marketers, entrepreneurs and educators do deals and talk, talk, talk about the future of education.

For the past two years, the mantra has been, "AI will transform everything."

This year? Not so much.

AI products and conversations were still everywhere, but the mood had shifted. There were a lot more questions - about evidence, about screen time backlash from parents, and about overwhelmed teachers. Superintendents were blunt. Their budgets are contracting and the era of "buy and try" is over.

"We're at a tipping point," said Erin Mote, CEO of InnovateEDU, a nonprofit organization.

Then, on the final day of the conference, Dan Meyer of Amplify, a curriculum company, dropped this newsletter bomb: "RIP Khanmigo & Edtech Industry Dreams of AI Tutors." The problem is that students aren't actually using AI tutors as much, or how, the way the industry hoped. They're not asking the kinds of questions that lead to learning.

A few other things I heard:

• Big ed tech companies are pivoting toward teacher training - trying to get teachers to drive adoption. Google is rolling out what it calls "snackable, stackable" ideas, bite-sized ways teachers can try AI in class without overhauling everything. Apple is expanding its networks of teachers who can coach colleagues on how to embed AI technology into their lessons.

• And industry research is starting to provide some answers. A small, exploratory study last year showed gains from an AI math tutor that gave instant feedback on practice problems. It used Google's AI under the hood to help correct students' misconceptions. But a much larger study - 1,600 students - just launched, with results expected this fall.

The big question now isn't only whether AI can help students if it is used properly. It's whether students will actually use it in ways that make a difference.

This story about the future of AI in the classroom was produced by The Hechinger Report, a nonprofit, independent news organization that covers education. Sign up for Proof Points and other Hechinger newsletters.

The post The future of AI in the classroom appeared first on The Hechinger Report.

The Hechinger Report

This story was originally published April 23, 2026 at 6:00 AM.