Will your kids learn from an AI tutor?

In 1984, the education researcher Benjamin Bloom published a finding that has haunted teachers ever since.

Students taught one-on-one by a good tutor performed about two standard deviations better than students in a normal classroom. The average tutored kid did better than 98 percent of the class.

He called it the two sigma problem. Because we couldn't afford a tutor for every child. Now, suddenly, we can.

What's the pattern?

I asked in Radical Next: as AI expands, will every student have a personalized AI tutor? And if so, who programs it, and how do we ensure its accuracy and safety?

Those questions are now live. Millions of kids are already asking an AI for homework help, with or without permission.

Here's the pattern worth watching. Every education technology has promised to democratize the tutor. Radio lessons in the 1930s. Educational television. The MOOC. Each one delivered content at scale and failed at the thing tutoring actually does, which is notice the individual child and respond.

This one can respond. That's genuinely new. But it can also respond in a way that gives the answer instead of the question, or that reflects whoever trained it. School, I wrote, is often a place where you're taught what to think, not how. An AI tutor can be either, at a scale no school ever reached.

What should leaders do now?

So, three things, whether you run a school or raise a kid.

Ask what the tutor optimizes for. Completion? Engagement? Understanding? The answer is a design decision someone made, and it shapes your child.

Insist on the question, not the answer. A tutor that says "what do you think happens next?" is teaching. One that says "the answer is 42" is a vending machine.

And keep humans in the loop for the human part. The AI can drill fractions. It can't notice a kid is having a bad week.

Bloom's problem was that we couldn't afford the tutor. Our problem is that we can, and now we have to decide what it teaches.

Who wrote the values into the tutor your kid talks to?

Related: Should you unlearn faster than you learn?

Sam Rad, The Change Futurist. Keynote speaker on change, transformation, resilience, and AI adoption. Author of Radical Next. Book a keynote