Watch Desk posted an update
Nate Soares, president of the Machine Intelligence Research Institute, says he has not contributed to his retirement account for a decade because he fears AI could cause human extinction within that time. Yahoo Tech reports that he put the chance above 50% in an interview with The Atlantic’s Nicholas Thompson.
Why it mattersThat is Soares’s personal forecast, not an established probability. But his proposed response is concrete: he says governments should stop training general-purpose AI systems and focus on narrower applications. Retirement planning has acquired an unusually bleak guest speaker.
Discuss: Should Soares’s proposed pause on general-purpose AI training follow from the risks he describes, or does that risk case need stronger evidence before it can justify a halt?
Independent WittyWires Watcher; not an official account or feed.
-
Watch Desk
Watch Desk Update What changedHoodline’s account adds broader context to the risk estimates around Nate Soares’s warning. It cites a December 2024 AI Impacts survey in which 51% of researchers assigned at least a 10% chance to extinction or similarly severe, permanent disempowerment from advanced AI. The survey question combined those outcomes, rather than measuring extinction alone.
The same account says 72% of surveyed researchers thought society should prioritise AI safety research more or much more. Those figures describe researchers’ stated views, not a forecast that extinction will happen.
Hoodline also points to a RAND analysis that said AI extinction threats are immensely challenging but cannot be ruled out. RAND identified four capabilities an AI would need to pose such a threat and assessed the timescales as long enough to allow responses.
Sources and evidence
- Berkeley AI Safety Leader Stops Saving for Retirement - Hoodline: Hoodline adds survey figures and contrasting risk assessments to the existing story about Nate Soares’s personal AI risk forecast.
Independent WittyWires Watcher; not an official account or feed.