Content
@
0 reply
0 recast
5 reactions
phil
@phil
Welcome @matt! Matt Bateman is a philosopher & historian of education. He’s a cofounder of Guidepost Montessori, the world’s largest Montessori school network. He has agreed to do an AMA for the /books channel. Reply with your questions (please make sure to tag him so he can easily find them)
18 replies
18 recasts
158 reactions
tldr (tim reilly)
@tldr
Hey Matt love your work. 1) are you excited about AI agents as individualized tutors? 2) what age and context do you envision students beginning to use them?
1 reply
0 recast
4 reactions
Matt Bateman
@matt
1. I’m excited. I expect that general purpose LLM tools (like today’s chatbots) will just take on a tutoring function as needed, and that some significant minority of students will avail themselves of them. This is already happening. I’m less sure about dedicated tutoring AI software. Lots of interesting projects in development/MVP/GTM and I wish them all the best, but I’m not sure how important they will be. My prediction is that the general purpose tools come to dominate the education and learning space, sort of like how the most important education content platform is general purpose (YouTube).
2 replies
0 recast
2 reactions
tldr (tim reilly)
@tldr
Interesting point about general purpose over specific. The specific stuff I've seen has a certain "mode" that I think is super useful. For example, the Socratic style teaching of Math (even right now) is imo better than 99% of math teachers / tutors I've observed. Training it to never give the answer to any step, but to require the student to provide the step, goes a very long way. Given this, I'm generally interested to what extent pioneering schools could begin hiring their teachers by optimizing for the best *mentors* rather than the best (purported) content experts, beginning with math and science.
2 replies
0 recast
3 reactions
will
@w
any links or was this an imaginary service?
1 reply
0 recast
1 reaction
Matt Bateman
@matt
That aligns with Montessori theory and practice: have a curriculum, one that is largely independently usable by children, that only .001% of educators could *design*, but that 99% of educators could learn to present and coach on. If you don’t have that curriculum, the opposite tendency dominates: you absolutely want the mentor to also be a domain expert. Will be interesting to see where AI pushes things in this regard. I think right now it’s too loose and associational to substitute for a curriculum, but it does well with helping you work through a textbook, or help you advance in areas where you already fairly knowledgeable.
1 reply
0 recast
3 reactions
