“the years it takes to build precision machinery.”
This one is an interesting question, I think. Right now, sure, this seems reasonable. But 1) it’s possible enough take-over-able precision machinery already exists, I don’t really know how to accurately evaluate that, especially for superhuman levels of ability to utilize what machinery exists, and 2) humans are building more and better precision machinery every day. So sure, a near-term fast-cognitive-takeoff AGI might have to bide its time while early plans bear fruit, but a later-term fast-cognitive-takeoff AGI (especially one in a world where people have been using its predecessors to accelerate R&D for a while) could easily come into being in a world where everything it needs for a big impact already exists.
“the years it takes to build precision machinery.”
This one is an interesting question, I think. Right now, sure, this seems reasonable. But 1) it’s possible enough take-over-able precision machinery already exists, I don’t really know how to accurately evaluate that, especially for superhuman levels of ability to utilize what machinery exists, and 2) humans are building more and better precision machinery every day. So sure, a near-term fast-cognitive-takeoff AGI might have to bide its time while early plans bear fruit, but a later-term fast-cognitive-takeoff AGI (especially one in a world where people have been using its predecessors to accelerate R&D for a while) could easily come into being in a world where everything it needs for a big impact already exists.