newsfilter.io

Katja Grace

Showing 11 of 1 transcripts.

  1. 80,000 Hours2h 12m

    #40 - Katja Grace on forecasting future technology & how much we should trust expert predictions

    Katja Grace, Keiran Harris, Rob Wiblin

    Led by Katja Grace, the three-year-old research organization AI Impacts employs empirical methods to forecast artificial intelligence timelines, finding that experts medianly predict high-level machine intelligence in 45 years while identifying significant biases in how these estimates are framed. Their analysis of compute trajectories reveals that the cost of hardware equivalent to the human brain has dropped dramatically, yet experts remain skeptical of sudden "intelligence explosions" due to historical precedents of incremental technological progress. Consequently, the team advocates for increased investment in AI safety research and the development of diverse forecasting skills to navigate the uncertainty surrounding future AI impacts.