Two camps of AI forecasters: extrapolating trends vs assuming growth will be paced

soumitrashukla9 · x · 2026-09-09

alexolegimas observes that AI forecasts split into two camps: those extrapolating current trajectories (higher, conditional forecasts) and those assuming capability growth is endogenous to decisions to slow down and "pace the frontier" — the open question is whether that premise holds.

He also pushes back on critics of "doomers": those closest to the technology have been the most accurate about capability growth. Separately, binarybits argues exponential revenue tells us little about AI capability, since every successful technology has a steep growth phase.

Original post →

More from AGI Musings

AGI Musings channel →