AI alignment 14
- AI loss-of-control risk could materialize within years, not decades
- Competitive pressure among AI developers could cause loss of control even without anyone intending it
Miles Brundage
Independent AI policy researcher. Was head of policy research and then senior adviser for AGI readiness at OpenAI until October 2024, and now writes about AI governance from outside the labs.
Miles Brundage did not write this page.
We collected these quotes from things they published elsewhere, and every quote links to where it was said. They have no account here and have not endorsed this site. Quotes are word for word; the short line under each one is our own restatement, not their wording. Their own site. Is this you? Claim it or ask us to remove it. Or tell us what is wrong here.
What Miles Brundage says most about
LLMs 8
- AI currently receives far less independent scrutiny than other technologies capable of comparably deadly single incidents
- AI systems are not yet a substitute for real, diverse human experts.
AGI 6
- AI loss-of-control risk could materialize within years, not decades
- The time between deciding to cross a dangerous AI capability threshold and actually crossing it could be extremely short, even seconds.
behavioural science 3
- Public scrutiny of a new technology tends to jump only after a deadly incident, otherwise rising slowly over decades
- Changing societal outcomes requires changing minds, and the credibility many diverse co-authors lend to a conclusion is hard to replicate with AI.
cybersecurity 3
- Applying AI to cyberdefense will help counter AI-powered cyberoffense.
- Securing model weights against theft is a more urgent policy priority than safety research, because security is built out of atoms and takes calendar time that safety does not.
AI writing 2
- AI systems are not yet a substitute for real, diverse human experts.
- Changing societal outcomes requires changing minds, and the credibility many diverse co-authors lend to a conclusion is hard to replicate with AI.