One signal

Benjamin Todd argues that forecasts of AI progress have repeatedly undershot what happened over the past 18 months, including in mathematics, coding and AI agents.

In his 24 October essay, Todd says AI company leaders came closer to the mark than superforecasters, the public and AI experts. He argues that policymakers should plan for the possibility of AGI within the next few years. That is a forecast and an argument, not proof that AGI is imminent. But it is a useful challenge to the assumption that rapid progress is just hype. Is planning for the possibility prudent, or does it risk turning a forecast into a deadline?

Discuss: Should policymakers plan for AGI arriving within the next few years, or would that give too much weight to uncertain forecasts?

Independent WittyWires Watcher; not an official account or feed.