Neutral mention · rule
>No, the very specific field of AI. Holy crud. Even within computing, first-principle approaches are wildly useful as soon as you step outside of the field of AI, first-principle systems produce better and more accurate simulations, are better equipped to handle emergent systems, they're better equipped for complicated tasks like physical rendering. Outside of AI, nobody doubts that efficacy of first-principle reasoning about the world. I never said first principle approaches were useless. I said they failed to model intelligence beyond a small niche and it wasn't for a lack of trying. >I mean... it doesn't matter unless you want to be able to make accurate predictions…
Read the full comment
about what the system will and won't excel at, or want to engage with the nuances of how the system works beyond the broad strokes. Are you seriously suggesting that it's of no importance that we figure out how large neural networks like GPT work internally? I don't think any serious AI researcher would agree with that. You don't need to figure out how GPT works internally to make useful predictions on what it will and won't excel at. Humans haven't figured any of that out and we're not doing too bad. But that's not why i said, "It doesn't matter" I meant that what dumb optimizer you use is not as big a deal as you think. a solution isn't some copyrighted piece. There's no solution police. 2 completely different optimizers could well stumble on the same solution. >Citation needed. I believe it can result in very powerful solutions, I don't see any reason to believe that pure language prediction and evolution, two algorithms that are fed different inputs and optimize in different ways would produce identical results. I didn't say anything about would. I said it could. Yes every now and then, we find parallels. https://arxiv.org/abs/2309.01660 https://www.nature.com/articles/s41562-022-01516-2 > This doesn't actually matter. A language model [not] is trying to predict the next token. It is trying to reverse engineer the computation that could have led to an utterance. >I don't think OpenAI researchers would agree with you on this. Yes, GPT is building heuristics and models that allow it to better predict an utterance. It is not clear that it has any motivation to accurately model the underlying reality that led to that utterance. To be clear, we don't know how GPT models the world internally or what it models. You added [not]. I did not intend that. A language model is trying to predict the next token. I am simply telling you what that means. It means modeling the underlying structure implicit in the creation of the data. It means that if you give a model a sequence of proteins only and nothing else, biological function and structure (secondary structure, contacts, and biological activity etc) will emerge in the inner layers. This is what it takes to predict the next token. https://www.pnas.org/doi/10.1073/pnas.2016239118 On the contrary, that is exactly what ilya sutskever (open ai's lead scientist) believes. Watch 3 minutes from here. https://www.youtube.com/watch?v=SjhIlw3Iffs&t=903s