3 ms·
Why we're unlikely to get artificial general intelligence anytime soon
- ggm 1y agoSeems caught up in the connectionist "just 1m more neuron-equivalents" and "its the wrong model" arguments. Which is trench warfare off to to one side of the essential agreement: Be it lack of scale, or irrelevant models, it isn't emerging and it won't emerge any time soon. Not a believer, so I tend to the "wrong model" side of things. I do think the orders of magnitude improvements in the quality of outcome are impressive but wishing for unicorns doesn't make them true, Mattel plastic not withstanding.
- baobun 1y agoAltman measures "achievement of AGI" in purely economic terms. Nothing to do with "consciousness" or "intelligence". Once global revenue tied to it hits x00M$ (or xB$ if they want to keep pushing and "adjust for inflation"), we have "AGI". So here it's entirely out of context. I think on the latest numbers they just need 25% returns on the big US center to get there.