DeepMind's Karvonen: Kurzweil's compute-extrapolation was '~all you need' to predict AGI

a_karvonen · x · 2026-09-06

In a debate with @phl43, akarvonen defends Kurzweil's 1999 AGI prediction: extrapolating compute growth and comparing it to the FLOPS of a human brain constituted an excellent argument. 'In retrospect, that's all you need to predict AGI.' phl43 counters that it is precisely the flawed argument he is criticizing.

Related event: Debate Erupts Over Whether Compute Extrapolation Can Predict AGI(6 posts)→

Original post →

More from AGI Musings

AGI Musings channel →