Will AI-driven differentiation of instruction based on individual error patterns substantially change education?

Andrew Feldman · Oct 6, 2025

Education will change substantially because AI can differentiate instruction based on the specific type of errors a student makes, which no one does today

Instruction has been essentially unchanged since Aristotle tutored Alexander; comparing a student's mistake patterns against thousands of others could route them to the remediation known to work for that pattern

56:45 Error pattern matching enables differentiated instruction impossible before

Severin Hacker · May 19, 2025

The best education has historically been reserved for the rich, and technology is the only way to give everyone access to one-on-one tutoring at the same efficacy

One-on-one human tutors work best but are too expensive for most people; software can deliver that experience at scale

5:01 Technology can democratize one on one tutoring quality at scale

Severin Hacker · May 19, 2025 · hedged

One-on-one tutors are effective mainly because they assess where you are, know where you need to go, and give calibrated feedback — which is why personalization is central to education's future

Like a tennis coach, a good tutor knows your current level and how you differ from others, and tailors feedback accordingly

9:05 Calibrated personalized feedback like a tutor is the future of education

Ethan Mollick · Jul 31, 2024

Active learning beats passive lecture-receiving, and flipped classrooms are a natural fit for AI — an adaptive AI tutor replaces the pre-class video and feeds what it learns into the classroom

Research supports active learning, and flipped classrooms already have early evidence in their favour; class time is where you can mess up in front of people, work in teams and get teacher help

48:43 Flipped classrooms are the natural structural fit for ai tutoring plus active learning

Ethan Mollick · Jul 31, 2024

'Order of magnitude improvement' is the wrong frame for education; any real improvement in the system would be a great outcome

Education is a complex system where every student has different talents, gaps and interests, so order-of-magnitude isn't measurable — even a 10% improvement is amazing

50:39 Any real improvement not order of magnitude is the right bar for education