Introduction
The tutor that refuses to give answers
An AI that writes homework can short-circuit learning. Salman Khan is building one that asks questions instead. The difference reaches into how schools teach, what grades measure, and why students still need difficult skills.
Rise of the AI tutor
Moving on leaves gaps that compound
A student who scores 80 percent usually advances with the class, carrying the missing 20 percent into harder material. Years later, difficulty with algebra may reflect unfinished learning about fractions rather than a lack of mathematical ability. Mastery learning gives students time to repair those foundations before advancing.
Benjamin Bloom’s 1984 work supplied Khan’s target: tutoring combined with mastery learning produced a two-standard-deviation improvement, moving an average student to roughly the 96th percentile. Providing that much human attention universally has been prohibitively expensive. Khan sees AI as a possible route to scale, while treating Bloom’s result as an aspiration rather than an established outcome for AI tutoring.
Takeaway
Keep the learning goal steady and let the time needed to reach it vary.