Yes, there are many misconceptions on "learning", but LLM can be useful. The approach that works for me is using Justin's skycak methods he mentions in his boo

Yes, there are many misconceptions on "learning", but LLM can be useful. The approach that works for me is using Justin's skycak methods he mentions in his books: https://www.justinmath.com/books/ https://www.justinmath.com/books/ Check the shorter "Advice on upskilling" or "The Math Academy way" for well researched approach. So what works for me - open a project in ChatGPT/Notebook LM - dump all the relevant and highly cited materials (textbooks, papers) - dump the advice on upskilling text or a short summary I've written for the LLM - create "Learning Goals", that contain what I want to learn, and how to estimate is my level good enough 1) Ask it to create a learning path from the materials, following the approach. Give that to an adversarial LLM for cross check. (just for sanity check) 2) Ask it to create an "entry test" to check what I do know and what I don't 3) Iterate step by step on each module/submodule from the learning path that intermingles the approach of: small theory step + small practical task + small test. Log what's missing/wrong in my dept log. Give the dept log at the end of the session to the LLM to incorporate/create another test/task. What I have found useful in this approach is that it will generate a lot of practical tests/tasks for me and it will explain a concept in many ways until I understand it. Also it finds some prerequisites I might miss, but based on my tests and debt log unexpected things I thought I understood surface. So with the limits of LLM and while building a mental map of the relevant parts it's usually enough to spot the hallucinations, but if you apply structured approaches these are minimal. And it's super good, because the number of practice tests and explanations is endless. The interfaces are a bit clunky, but current multimodal LLMs are ok with images or even hand writing. I will recommend that structured approach.