Not the original commenter, but I think LLMs (GPT in particular) are an incredibly powerful study / research tool. They allow you to answer questions you'd otherwise take weeks to have answered or not be able to feasibly answer at all. With your questions answered you can move on instead of getting stuck.
This also works at a meta level. E.g. you can use them to research how to improve your learning or thinking, and boost yourself to a superhuman level.
🎯 .. LLMs are very adept at matching the intent of what you're looking for to actually useful information. There is less "skill" required to engineer "the perfect search term". You can ask the question the way you would ask it and it will generally get you close to the target within one or two volleys.
As I covered in this one minute section from @jsonbits meetup - the reason for this is because vectorization allows for the "quantization of language". Meaning you can ask the same question 40 ways with different vernacular or perhaps even different languages and still get a similar answer that's meaningful & useful! It's really a revolution in flexibility computational systems above everything else.
reply