Post
That study about 95 percent of enterprise AI pilots failing shouldn't be surprising to anyone who has studied the diffusion of past general-purpose technologies.
English
@binarybits @karpathy This was actually the part of the interview that I had the most issues with. I don't understand how you can think we are 10 years away from AGI and think we are going to stay at 2% GDP growth. I think "AI as normal tech" is reasonable, but not compatible with AGI in 10 years.
English
@binarybits @karpathy I mean, he also said that he worries AI will disempower humanity and that AI will build Dyson spheres in the same interview.
I think he just believes this will happen smoothly over 10-20 years, rather than discontinuously in 2-3 years.
English
@binarybits @karpathy hasn't AI already had a bigger GDP impact? I keep hearing "if you subtract datacenter capex, we're in a recession"
English
@binarybits @karpathy I think both could be true that we’ve been on the exponential, but also that the exponential is about to become abundantly obvious and radically transformative
Space Man@starsailing11
@haider1 But: maybe it’s just the oddity of exponentials, and it’s saying the same thing: If you put a single drop of water into an Olympic pool, then doubled it every minute, the pool fills in 50 minutes. However it’s HALF full at 49 minutes, & 1/1000th full (unnoticeable!) at 40mins
English