2 ms·
> "if you're expecting AI, LLMs don't cut it". And I think he's basically right on that count. This is one of those comments whose truth value depends entirely
by antonvs 8mo ago
> "if you're expecting AI, LLMs don't cut it". And I think he's basically right on that count.
This is one of those comments whose truth value depends entirely on a constantly shifting definition of “AI”.
The ability of modern models to functionally understand, answer questions, and make recommendations about software codebases is superhuman at this point, relative to most human software developers. What is that, if not artificial intelligence?
Perhaps you’re thinking of something more like AGI, but even there the terminology is loaded and ambiguous. The models are general enough to answer questions well on a vast range of subjects, and they exhibit understanding (again, functionally speaking this is true - whether someone wants to call them stochastic parrots is beside the point.) The appellation of “intelligence” applies just as well as in the coding case, it’s artificial, and it’s general.
> a worthwhile point to make.
I disagree. Without clear, justified definitions, it’s an incoherent, poorly specified point that seems to be driven by a desire to maintain a specific conclusion regardless of the evidence.
- albatross79 8mo ago[flagged]
- logicprog 8mo agoI'm not sure what you mean by "everyone knew what AI meant"? It's always been an extremely vague term, since its inception — and LLMs do, actually, enable computers to do things I, at least, would've called basically AGI in 2020.
- albatross79 8mo agoThere's nothing vague about it. Intelligence is understanding. I think you're just motivated to be confused because you want to fit this thing into that box and it doesn't really fit. So maybe output is an that matters, are we auto completers, can you be intelligent without knowing, etc etc.
- logicprog 8mo agoI'm sorry, but very many systems have been called AI that don't have "understanding," and been very useful without that. I'm not sure when that was ever part of the definition? Even Wikipedia gives a definition similar to my own, and I think Karen Hao makes a very good historical case that the definition of AI has never been super clear in Empires of AI, and she's no fan of LLMs (or the field of AI in general, it seems), that there never really has particularly been a clear definition.
- antonvs 8mo ago> Everyone knew what AI meant before LLMs came along Please do tell. > Why is everyone on the ground floor trying to defend what the guy in the penthouse is telling the investing public? I'm talking about observable capabilities of the technology. There are facts of the matter there that have nothing to do with what tech CEOs say. Pointing out the reality of the situation is not "defending" anyone. If you take a "team sports" approach to reality where truth is a function of which side you're on, then you're not going to able to participate meaningfully in rational discussion.
- albatross79 8mo agoThere's rational discussion and then there's motivated obfuscation. If your definition of AI is so loose that it applies to a system that can't count objects reliabily, then what is there to discuss? Is a broken clock intelligent because it tells the time correctly twice a day?