@TomDavidsonX @willmacaskill The best econ models that we have (eg GATE) that incorporate AI automation only get to 7 OOMs of output growth in twenty years. And that one is quite idealised.
12 OOMs is *a lot*. Not impossible, but enough of a different number that it should give you pause.
Tom Davidson@TomDavidsonXFyi @Jsevillamol @willmacaskill
More thoughts: 1. I think these better models screen off the historical prior, which would also have ruled out the 70 year merely exponential growth we are living through today. 2. It doesn't have to be energy that bottlenecks growth. Other non-accumulable resources or regulation could slow down growth. 3. A 7 year old reversion to merely exponential growth after AI as we solve the next bottlenecks to growth wouldn't be unbelievable, and would plausibly put a Dyson sphere behind schedule, even with much faster efficiency growth than predicted by models.
Jaime Sevilla@Jsevillamol@TomDavidsonX @willmacaskill The best econ models that we have (eg GATE) that incorporate AI automation only get to 7 OOMs of output growth in twenty years. And that one is quite idealised. 12 OOMs is *a lot*. Not impossible, but enough of a different number that it should give you pause.