Post

From some 27 years of designing courses, I think human teachers should be focused on where the design of the teaching implicitly teaches the scaffolding for deeper understanding the illustrating principles. The LLM response to student prompt doesn't come from a place of that understanding.

Absolutely, I haven't found LLMs to be great at teaching mental models. They also don't empathise with the learner or help with the "you don't know what you don't know" problem. But when you've reached an intermediate level there's potentially a lot of value in using them to learn.

I think lurking in there somewhere is a distinction between reinforcement of existing knowledge, where a person reflectively knows their own weaknesses and can articulate them, and learning a new area, where the person does not have the context of the size and scope of their weaknesses