This is my running analogy for AI as a general-purpose technology:
If this analogy isn’t far off the mark, it gives a reference class for thinking about replacement rates.
Vaclav Smil’s calculations of replacement levels of animate “prime movers” (human labor, horses, oxen) were gradually replaced animate movers, increasing their share by a bit less than 10% a decade.
So, what’s the reason for expecting AI to replace human problem-solvers faster?
There’s a couple of pieces to the answer, all of which I plan to elaborate on in the future, but two major pieces are:
- AI automation can potentially relieve bottlenecks on growth in human production:
- Research and development
- Design of hardware and software systems
- Task decomposition
- AI is going to operate at a much higher speed than humans.