I'd like to see more detailed analyses of how/how quickly massive job automation would happen.
I think there's a risk of the planning fallacy - that you overrate the speed because many obstacles are non-obvious.
Cf the fairly slow self-driving car rollout despite the tech being pretty good now.
I think there's a risk of the planning fallacy - that you overrate the speed because many obstacles are non-obvious.
Cf the fairly slow self-driving car rollout despite the tech being pretty good now.
Comments
I also think there's a risk of underrating the speed because the jump from "not very good" to "superhuman" is comparatively tiny. Also because the moment agents are good enough they can scale massively (unlike cars).
Also, long back-and-forth w Matt Clancy ending here: https://x.com/steve47285/status/1796546304811868171
(This is about future AI not Jan 2025 AI)