6 ms·
I don't see how this gets us (much) closer to general AI. Where is the reasoning?
by tpmx 4y ago
I don't see how this gets us (much) closer to general AI. Where is the reasoning?
- _joel 4y agoPerhaps the confluence of NLP and something generative?
- astrange 4y agoThat doesn’t even lead in the direction of an AGI. The larger and more expensive a model is the less like an “AGI” it is - an independent agent would be able to learn online for free, not need millions in TPU credits to learn what color an apple is.
- SemanticStrengh 4y agoYes metaculus mostly bet a magic number based on perhaps and tbh why not, the interaction of NLP and vision is mysterious and has potential. However those magic numbers should still be considered magic numbers. I agree that in 2040 the interactions will have extensively been studied though but the conclusion of wether we czn go much further on cross-models synergies is totally unknown or pessimist.
- quirino 4y agoI think this serves at least as a clear demonstration of how advanced the current state of AI is. I had played with GPT-3 and that was very impressive but I couldn't even dream something as good as D-ALLE 2 was already possible.
- 6gvONxR4sf7o 4y agoBig pretrained models are good enough now that we can pipe them together in really cool ways and our representations of text and images seem to capture what we “mean.”
- tpmx 4y agoYeah, it seems like it. But it's still just complicated statistical models. Again, where is the reasoning?
- 6gvONxR4sf7o 4y agoI don’t care whether it reasons its way from “3 teddy bears below 7 flamingos” to a picture of that or if it gets there some other way. But also, some of the magic in having good enough pretrained representations is that you don’t need to train them further for downstream tasks, which means non-differentiable tasks like logic could soon become more tenable.
- renewiltord 4y agoA belief oft shared is that sufficiently complicated statistical models are indistinguishable from reasoning.
- marvin 4y agoI still think we're missing some fundamental insights on how layered planning/forecasting/deducting/reasoning works, and that figuring this out will be necessary in order to create AI that we could say "reasons". But with the recent advances/demonstrations, it seems more likely today than in 2019 that our current computational resources are sufficient to perform magnificantly spooky stuff if they're used correctly. They are doing that already already, and that's without deliberately making the software do anything except draw from a vast pool of examples. I think it's reasonable, based on this, to update one's expectations of what we'd be able to do if we figured out ways of doing things that aren't based on first seeing a hundred million examples of what we want the computer to do. Things that do this can obviously exist, we are living examples. Does figuring it out seem likely to be many decades away?
- tpmx 4y agoThat's a well-balanced response that I can agree with. I'm an not AGI-skeptic. I'm just a bit skeptical that the topic of this thread is the path forward. It seems to me like an exotic detour. And, of course intelligence isn't magic. We're producing new intelligent entities at rate of a about ~5 per second globally, every day. > Does figuring it out seem likely to be many decades away? 1-7?