Trend extrapolation is more reasonable than invoking something that hasn’t happened at all yet, and then claiming, “When this happens, it will become an unstoppable trend.”
It would be more reasonable to use trend-extrapolation if it was a field where you would necessarily be able to discern a trend. Yudkowsky argues there could be sharp discontinuities. Personally I don’t really feel qualified to have a strong opinion, and I would not be able to discern a trend even if it exists.
Trend extrapolation is more reasonable than invoking something that hasn’t happened at all yet, and then claiming, “When this happens, it will become an unstoppable trend.”
It would be more reasonable to use trend-extrapolation if it was a field where you would necessarily be able to discern a trend. Yudkowsky argues there could be sharp discontinuities. Personally I don’t really feel qualified to have a strong opinion, and I would not be able to discern a trend even if it exists.