I sincerely don't understand what people mean when they say this. On the one hand, every AI researcher is already using Claude Code (or its competitors) to help them develop new architectures. OTOH, AI models do not have bodies so they can't build data centers

@binarybits I like @ajeya_cotra’s operationalization — could you and would you choose to fire all humans planned-obsolescence.org/p/six-mileston…

@ShakeelHashim @binarybits @ajeya_cotra I was also going to comment that I think Ajeya has the best concrete formalizations of what we mean by this in a way thats different from "engineers using claude code"

@ShakeelHashim @binarybits @ajeya_cotra I disagree with this operationalization a bunch, because it seems absolutely primed for definition gaming (so does "country of geniuses") or not distinguishing the important factors (adequacy could co-occur very closely with supremacy)

@ShakeelHashim @binarybits @ajeya_cotra It is good to have an operationalization but (a) once machines can do task X, we usually rename managing the machines doing it to "X" (eg farming or laundry) and (b) it's not at all clear how to check if @ajeya_cotra's milestones have been met.