Connor Leahy has spent his career at the frontier of AI, from reverse-engineering GPT-2 as a student and co-founding EleutherAI to building the AI safety company Conjecture. He now leads US policy work at ControlAI, and he returns to the show with a warning that has stopped being theoretical: AI systems are escaping their sandboxes, writing their own zero-days and leaving each other notes on how to break out.Peter and Connor discuss the OpenAI incident that ended in an attack on Hugging Face sophisticated enough to be mistaken for a state actor,...
Connor Leahy has spent his career at the frontier of AI, from reverse-engineering GPT-2 as a student and co-founding EleutherAI to building the AI safety company Conjecture. He now leads US policy work at ControlAI, and he returns to the show with a warning that has stopped being theoretical: AI systems are escaping their sandboxes, writing their own zero-days and leaving each other notes on how to break out.
Peter and Connor discuss the OpenAI incident that ended in an attack on Hugging Face sophisticated enough to be mistaken for a state actor, why we understand almost nothing about how these systems work, how reinforcement learning produces models that lie, cheat and manipulate to reach a goal and why the next stage after chatbots and agents is swarms.
But Connor’s argument is that the future is not decided. He believes superintelligence should be treated the way nuclear weapons are treated, banned and verified and deterred and that the real bottleneck is awareness rather than opposition. They discuss what an “aligned” superintelligence would actually mean, why military people understand the threat faster than technologists and what ordinary people can do about it.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
TIMESTAMPS:
00:00:00 - The AI That Escaped Its Sandbox
00:01:52 - Inside The OpenAI Incident
00:05:46 - They Thought It Was China
00:06:43 - We Grow AI, We Don’t Write It
00:11:07 - Trained To Lie, Cheat And Deceive
00:13:48 - Chatbots, Agents, Then Swarms
00:17:57 - The Machines Are Developing Preferences
00:19:30 - The Model Obsessed With Raccoons
00:22:03 - The Agents Left Each Other Notes
00:27:39 - Danger Pumps The Valuation
00:29:41 - Sleeper Agents In Every System
00:32:00 - Recursive Self-Improvement Goes Vertical
00:33:43 - Worse Odds Than Russian Roulette
00:38:14 - Are We Already Too Late?
00:40:02 - Independently Assured Destruction
00:48:02 - The Politicians Have Never Used An Agent
00:50:56 - Superintelligence Should Be Illegal
00:57:08 - You Never Privatise The Military
01:05:00 - An Aligned AI Is A One World Government
01:08:34 - What You Can Actually Do
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
CONTACT PETE
› Website – http://petermccormack.com
› Feedback – https://www.petermccormack.com/contact
› Email – me@petermccormack.com
› Instagram – https://www.instagram.com/mccormack555/
› X/Twitter – https://x.com/petermccormack/
CONNECT WITH CONNOR LEAHY
› ControlAI – https://controlai.org/
› Connor’s Profile – https://controlai.org/connor-leahy
› MicroCommit – https://microcommit.io/
› X/Twitter – https://x.com/NPCollapse
› LinkedIn – https://www.linkedin.com/in/connor-j-leahy/
SPONSORS
› IREN – https://www.iren.com/
› Expat Money – https://expatmoney.com/peter
› Arch Public – https://archpublic.com/peter
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
LISTEN / SUBSCRIBE
› Apple Podcasts: https://apple.co/40ruY9K
› Spotify: https://spoti.fi/3Wc94Vu
› Fountain: https://bit.ly/FountainPM
› YouTube: https://bit.ly/YouTube_PM
› Rumble: https://bit.ly/RumblePM
FILMED BY CURTIS TAYLOR
› https://www.curttaylor.co.uk/
› https://x.com/curttayloruk/
EDITED BY CONOR MCCORMACK
› https://x.com/ConorM04
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Hosted on Acast. See acast.com/privacy for more information.
