The AIs have their own biases and shortcomings, and it can be very hard to correct for them. It can feel very good to feel like you can understand lots of things with them, but please be cautious! Perhaps take the time to test a fanciful/cranky belief you have, but one you know to be false, and try to get the LLM to convince you of it, as suggested here: https://www.lesswrong.com/posts/iuMpzeLfRhsH6ntRf/kbear-s-shortform?commentId=wrWAipieJSSr23aMT It may give you some evidence about how much you should change your beliefs about other things the AI tells you.
Hi, and welcome! These sorts of spaces—rationalist, or rationalist-adjacent, can be a breath of fresh air for discourse, I totally agree!
Re: using LLMs for knowledge generation, I would caution that it’s quite possible to become disconnected from reality, like mentioned here: https://www.lesswrong.com/posts/rarcxjGp47dcHftCP/your-llm-assisted-scientific-breakthrough-probably-isn-t
The AIs have their own biases and shortcomings, and it can be very hard to correct for them. It can feel very good to feel like you can understand lots of things with them, but please be cautious! Perhaps take the time to test a fanciful/cranky belief you have, but one you know to be false, and try to get the LLM to convince you of it, as suggested here: https://www.lesswrong.com/posts/iuMpzeLfRhsH6ntRf/kbear-s-shortform?commentId=wrWAipieJSSr23aMT It may give you some evidence about how much you should change your beliefs about other things the AI tells you.