I wrote a LessWrong article that tries to estimate doubling time for a self-reproducing robot. A critical step is that smaller robots are faster. Most manufacturing processes scale such that they get N times faster as they get N times smaller. I picked N=4, for reasons explained in the article. I concluded the doubling time is five weeks. So the time to a billion robots is on the order of five years.
Even if your goal is a human-size robot, you’re better off building small robots to build it, since they work faster. I assumed fairly clumsy hardware, but software comparable to a human machinist in cleverness.
The bottlenecking effect of size can be significantly reduced by eg replacing EM links with optical. We can’t do this yet, but something not very far above our intelligence might. It will probably find other ways to start taking apart the bottlenecks that seem static to us, being able to see further than we can. If the software starts out comparable to a human machinist and self-improves, surely your foresight horizons on its exact technical bottlenecks will quickly expire.
None of that excuses the exclusive cultural fixation on big bulky humanoids for tasks where a wheeled spider would obviously be better suited.
[ Re OP, I just can’t visualize converting that many car factories that fast when IIRC Elon had trouble simply convincing manufacturers to make Tesla doors an unconventional shape. Maybe China could make a billion humanoids in 20 years? But by then we have non-humanoid ASI. ]
I wrote a LessWrong article that tries to estimate doubling time for a self-reproducing robot. A critical step is that smaller robots are faster. Most manufacturing processes scale such that they get N times faster as they get N times smaller. I picked N=4, for reasons explained in the article. I concluded the doubling time is five weeks. So the time to a billion robots is on the order of five years.
Even if your goal is a human-size robot, you’re better off building small robots to build it, since they work faster. I assumed fairly clumsy hardware, but software comparable to a human machinist in cleverness.
The bottlenecking effect of size can be significantly reduced by eg replacing EM links with optical. We can’t do this yet, but something not very far above our intelligence might. It will probably find other ways to start taking apart the bottlenecks that seem static to us, being able to see further than we can. If the software starts out comparable to a human machinist and self-improves, surely your foresight horizons on its exact technical bottlenecks will quickly expire.
None of that excuses the exclusive cultural fixation on big bulky humanoids for tasks where a wheeled spider would obviously be better suited.
[ Re OP, I just can’t visualize converting that many car factories that fast when IIRC Elon had trouble simply convincing manufacturers to make Tesla doors an unconventional shape. Maybe China could make a billion humanoids in 20 years? But by then we have non-humanoid ASI. ]