AGI will be slow takeoff because:
1. Fast takeoff implies that AGI crosses the threshold of human
intelligence and starts improving itself. But no such threshold
exists. It depends on how you measure intelligence. Computers are
already a billion times smarter on tests of arithmetic and short term
Ok, we have been in para-singularity mode for about a year now. What are
the next steps?
I see two possibilities:
A. AGI cometh. AGI is solved in an unambiguous way.
B. We enter a "takeoff" scenario where humans are removed from the
upgrade cycle of AI hardware and software. We would start g
