3 ms·

You know what's even more over-hyped is GPT-4. Sam Altman recently stated that many people will be disappointed because of all of the rumours that it's AGI.

by ilaksh 4y ago

waynesonfire 4y ago
now that's starting to tread hype territory. AGI is a COMPLETELY different beast.
HarHarVeryFunny 4y ago
Yeah, it's a useful scaling experiment given the improvement from GPT-2 to GPT-3.5 (base of ChatGPT), but I'd have to guess the "emergent capability" curve is flattening out at this point. ChatGPT already does well when essentially regurgitating factually correct information, but the obvious weakness of an LLM is that when combining sources its just combining them linguistically so it produces fine sounding streams of often factually incorrect bullshit. It's hard to imagine how GPT-4 will be much different unless there have been some fundamental extensions to the model architecture, or changes to the training regime. Presumably it'll do better in "search engine mode" where people are hoping it'll "understand" the subtleties of their questions in selecting what to generate, but hard to see how it'll avoid the "bullshit chained deduction" problem unless they've added a way for it to learn from it's own mistakes. In terms of perception/disappointment, I guess it depends on individual expectations. If you want GPT-4 to be a better "search engine" than GPT-3, then you may be happy. If you expect GPT-4 to be "more intelligent" than GPT-3, then I'm guessing not so much. Let's see! I've a hunch it may do well as the basis of an improved Copilot since this seems to be an easier subdomain.
- postalrat 4y ago
  AGI is still AGI even if it's often wrong.
  - HarHarVeryFunny 4y ago
    Well, there's a bar to be called AGI vs AI, or AI vs ML for that matter. Any program that only gets things right by being lucky enough to come up with the right "search key" for a correct answer isn't AI in my book. YMMV.
- HarHarVeryFunny 4y ago
  On consideration, maybe "combining them linguistically" is a bit too literal. Of course that is what a language model does, but that ignores that an extremely good language model such as GPT-3+ will have had to have learned a pretty good world model in order to do a good job of next-word prediction, to the extent that regarding this as merely continuation statistics (while true) is a bit too dismissive of what it's actually capable of. So, perhaps the scale upgrade from GPT-3 to GPT-4 will improve it's world model enough that it does a better job of combining sources using a consistent bias/POV rather then happily combining true and false sources as it currently does.

You know what's even more over-hyped is GPT-4. Sam Altman recently stated that many people will be disappointed because of all of the rumours that it's AGI.

by ilaksh 4y ago

You know what's even more over-hyped is GPT-4. Sam Altman recently stated that many people will be disappointed because of all of the rumours that it's AGI.

waynesonfire 4y ago

now that's starting to tread hype territory. AGI is a COMPLETELY different beast.

HarHarVeryFunny 4y ago

Yeah, it's a useful scaling experiment given the improvement from GPT-2 to GPT-3.5 (base of ChatGPT), but I'd have to guess the "emergent capability" curve is flattening out at this point. ChatGPT already does well when essentially regurgitating factually correct information, but the obvious weakness of an LLM is that when combining sources its just combining them linguistically so it produces fine sounding streams of often factually incorrect bullshit. It's hard to imagine how GPT-4 will be much different unless there have been some fundamental extensions to the model architecture, or changes to the training regime. Presumably it'll do better in "search engine mode" where people are hoping it'll "understand" the subtleties of their questions in selecting what to generate, but hard to see how it'll avoid the "bullshit chained deduction" problem unless they've added a way for it to learn from it's own mistakes. In terms of perception/disappointment, I guess it depends on individual expectations. If you want GPT-4 to be a better "search engine" than GPT-3, then you may be happy. If you expect GPT-4 to be "more intelligent" than GPT-3, then I'm guessing not so much. Let's see! I've a hunch it may do well as the basis of an improved Copilot since this seems to be an easier subdomain.

postalrat 4y ago
AGI is still AGI even if it's often wrong.
- HarHarVeryFunny 4y ago
  Well, there's a bar to be called AGI vs AI, or AI vs ML for that matter. Any program that only gets things right by being lucky enough to come up with the right "search key" for a correct answer isn't AI in my book. YMMV.
HarHarVeryFunny 4y ago
On consideration, maybe "combining them linguistically" is a bit too literal. Of course that is what a language model does, but that ignores that an extremely good language model such as GPT-3+ will have had to have learned a pretty good world model in order to do a good job of next-word prediction, to the extent that regarding this as merely continuation statistics (while true) is a bit too dismissive of what it's actually capable of. So, perhaps the scale upgrade from GPT-3 to GPT-4 will improve it's world model enough that it does a better job of combining sources using a consistent bias/POV rather then happily combining true and false sources as it currently does.