LLMs present a tremendous opportunity to scale up human learning of well known topics by providing a tutor for all. LLMs are an answer to many of the problems faced by learners such as where to start or how to go deeper on a particular topic. What excites me most about AI (well, LLMs specifically) is the ability to compress and synthesize knowledge. Rather than a student having to search for an answer, they need only ask. LLMs have a reputation for hallucination, and I don't disagree with that reputation at all, however, in my experience they perform admirably when it comes to _regurgitating well known information_. I'm talking computer science, physics, math, etc. Generally, STEM topics. There's a category of information that a huge number of humans already know, in absolute terms, but which is still very useful to learn. Lots of technical knowledge falls under this heading. # Apprenticeship With all that being said, the LLM as a learning companion still has its limits. It was trained on public information after all, so it likely _doesn't know_ about the intricacies of the real world in various professions. For deep knowledge in some field, apprenticeship is still going to have tremendous value. For this reason I don't think human apprenticeship is going anywhere. AI isn't going to fully replace other elements of a human's education, but rather, server as a compliment.