Ask HN: Is there a "Slow but thoughtful" LLM?
Looking for something where I can give it a topic or area of interest, and it does an extremely comprehensive deep dive.
The ideal version would be that I can give it a topic and a level of understanding (ELI5, ELI10, ELI25) and it would summarize and organize all human knowledge on that topic and provide it to me. I would pay $X00 for this and would wait even 24-48 hours... > The ideal version would be that I can give it a topic and a level of understanding (ELI5, ELI10, ELI25) and it would summarize and organize all human knowledge on that topic and provide it to me. I would pay $X00 for this and would wait even 24-48 hours... That exists, its called a textbook. https://www.lesswrong.com/posts/xg3hXCYQPJkwHyik2/the-best-t... “ For years, my self-education was stupid and wasteful. I learned by consuming blog posts, Wikipedia articles, classic texts, podcast episodes, popular books, video lectures, peer-reviewed papers, Teaching Company courses, and Cliff's Notes. How inefficient! I've since discovered that textbooks are usually the quickest and best way to learn new material. That's what they are designed to be, after all.” Unfortunately, there is no textbook on something very particular like "The CPAP Market in the US" - which is what I want to understand currently. Also textbooks frequently miss out on the level of understanding part of this. There is no textbook that explains quantum mechanics for smart 15 year olds. > There is no textbook that explains quantum mechanics for smart 15 year olds That's because you need a level of mathematical maturity to understand modern physics and QM that 99% of 15 years old haven't done. Maybe you were trying to prove a point but that's a bad example, pedagogically speaking. Point being by the time you're ready to meaninfully learn QM you don't need to dumb it down to ELI15. Anything else is pop-science. IF you really want to learn this stuff you can start here: https://goodtheorist.science/. Note the first steps are learn enough math so you can 'speak the language of physics'. > The CPAP Market in the US Have you done a study like this before?, I'm not sure if you know but: Again an LLM is not going to be able to know this because the data comes from private primary research, which you would either gather yourself, sift through trade reports, or engage a consulting or market research firm to go find out. This is firmly beyond current LLM's ability to synthesize, since the data is literally: a) private, paywalled, expensive to procure, b) extremely temporal, in that the market landscape changes year to year or even faster. You would start somewhere like https://www.bain.com/insights/industry-insights/healthcare-i... and then need to go buy things like https://www.premiummarketinsights.com/reports-tip/north-amer... and then trawl https://www.linkedin.com/pulse/global-respiratory-care-devic... for experts who you might need to pay to interview. Sure some of the companies are publicly traded, in which case you can go look at 10k forms etc., but the sources of knowledge are private, expensive, and rapidly changing for that kind of question. No, not really. The biggest AI models can currently run at near-realtime, so you're not missing out on much with the immediacy of it. The "premium" version of AI is doing your own research. AI will be wrong whether you run it fast or slow, so you've got to take that into account from square one.