factual
AI alignment progress is in the conversation
Real progress on AI risk has been in the conversation rather than the technical solution: after Bostrom's Superintelligence, prominent figures like Demis Hassabis and Stuart Russell came to treat AI safety as serious, with broad agreement that the threat is decades away and premature regulation would be a mistake, while scientists should devote more than a tiny fraction of effort to alignment—yet the technical alignment work itself still has a long way to go.
factualpending
Speaker
Toby OrdEvidence Quote
“i think we've made a lot of progress on the conversation around it ... but when it comes to the particular approaches for alignment ... it does look like there's a long way to go”
Created: 6/14/2026, 1:52:06 AM
Your notes
Loading notes...