In 1984, the educational psychologist Benjamin Bloom published a result that has haunted schools ever since. Students taught one-to-one by a tutor did about two standard deviations better than students taught in a normal class. The average tutored student outperformed 98% of the class.
98%
Share of classroom students the average tutored student outperformed.
Source: Benjamin Bloom, Educational Researcher, 1984
Bloom called it the 2 sigma problem. The problem was never the method. It was that a personal tutor for every child is impossible to pay for — there aren't enough people, or hours.
What changed#
An AI assistant can now do much of what a patient tutor does: explain an idea three different ways, ask you a question, wait for your answer, and tell you where you went wrong. It is uneven — sometimes wrong, sometimes brilliant — so it works best when you use it the way a good student uses a tutor, not the way a lazy student uses an answer key.
Try it tonight#
Pick something you or your child is learning, open any AI assistant, and paste this:
I want to learn [the topic]. Don't give me the answers. Ask me one question at a time, wait for my answer, then tell me what I got right, what I got wrong and why. Start easy and get harder. Every few questions, check that I can explain the idea in my own words.
Three habits make it work:
- Answer before you read. The learning happens when you try, not when you read the explanation.
- Ask it to show its working. If it states a fact that matters, ask where it comes from — and check the important ones in a textbook or a source you trust.
- Stop when you can teach it. If you can explain the idea to someone else without notes, you've got it.
Why it matters for every life#
For the first time, the method Bloom measured can cost almost nothing. Whether it reaches every child depends less on the model than on the phone, the language and the trust around it — and on someone showing them how to ask.