Testing limits: models detect evaluations

Stephen Adler is a former safety researcher at OpenAI. He joins the podcast to discuss how to govern increasingly capable AI systems. The conversation covers competitive races between AI companies, limits of current testing and alignment, mental health harms from chatbots, economic shifts from AI labor, and what international rules and audits might be needed before training superintelligent models. 

LINKS:

• Steven Adler's Substack: https://stevenadler.substack.com

CHAPTERS:
(00:00) Episode Preview
(01:00) Race Dynamics And Safety
(18:03) Chatbots And Mental Health
(30:42) Models Outsmart Safety Tests
(41:01) AI Swarms And Work
(54:21) Human Bottlenecks And Oversight
(01:06:23) Animals And Superintelligence
(01:19:24) Safety Capabilities And Governance

PRODUCED BY:

https://aipodcast.ing

SOCIAL LINKS:

Website: https://podcast.futureoflife.org

Twitter (FLI): https://x.com/FLI_org

Twitter (Gus): https://x.com/gusdocker

LinkedIn: https://www.linkedin.com/company/future-of-life-institute/

YouTube: https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/

Apple: https://geo.itunes.apple.com/us/podcast/id1170991978

Spotify: https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP