If you aren't aware: (high-parameter-count) LLMs can be used pretty reliably to teach yourself things.
LLM base models "know things" to about the same degree that the Internet itself "knows" those things. For well-understood topics — i.e. subjects where the Internet contains all sorts of open-source textbooks and treatments of the subject — LLMs really do "know their shit": they won't hallucinate, they will correct you when you're misunderstanding the subject, they will calibrate to your own degree of expertise on the subject, they will make valid analogies between domains, etc.
Because of this, you can use an LLM as an infinitely-patient tutor, to learn-through-conversation any (again, well-understood) topic you want — and especially, to shore up any holes in your understanding.
(I wouldn't recommend relying solely on the LLM — but I've found "ChatGPT in one tab, Wikipedia open in another, switching back and forth" to be a very useful learning mode.)
See this much-longer rambling https://news.ycombinator.com/item?id=43797121 for details on why exactly this can be better (sometimes) than just reading one of those open-source textbooks.
