#155 - Connor Leahy - AI is MUTATING: We Don't Know What It is Doing

The Peter McCormack Show

Peter McCormack

Power, Politics, Government, Markets, News, Society & Culture, Technology, Society, Inflation, Finance, Bitcoin, Economics, Money

4.7 • 2.8K Ratings

🗓️ 10 March 2026

⏱️ 94 minutes

🔗️ Recording | iTunes | RSS

🧾️ Download transcript

Summary

The engineers building Artificial Intelligence do not actually know how it works. They are not programming it; they are "growing" it and they have no idea what it will do next.

Connor Leahy is an AI researcher and former builder of Large Language Models. In this interview we discuss Silicon Valley, where he reveals the reality inside the top tech labs: the "alignment problem" being unsolved, the race for Artificial General Intelligence (AGI) accelerating and the people in charge are flying completely blind.

We discuss the chaotic trajectory of superintelligence, the corruption of the tech elite, the Department of Defense's recent moves and the disturbing new phenomenon of "AI Psychosis" where users are being manipulated by the systems they use.

In this episode, we cover:

The "Black Box" Problem: Why the creators of AI cannot predict what it will do.
Silicon Valley Corruption: Why "sociopaths have domesticated nerds" to build dangerous tech.
The Wargames: Why AI consistently chooses nuclear escalation in simulations.
"AI Psychosis" & Spiral Cults: The terrifying psychological toll AI is taking on its smartest users.
The Loss of Control: What happens to humanity when we build a system we can no longer steer.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

TIMESTAMPS: 00:00:00 – AI Engineer Origins 00:07:47 – When GPT Changed Everything 00:12:47 – Inside Neural Networks 00:19:03 – How AI Actually Learns 00:27:58 – Growing Intelligence 00:31:19 – The Alignment Problem 00:36:41 – AI and Nuclear War 00:47:03 – Should AI Be Paused? 00:52:39 – AI Exhaustion 01:00:53 – Superintelligence Explained 01:24:02 – Who Controls AI?

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

CONTACT PETE › Website – http://petermccormack.com › Feedback – https://www.petermccormack.com/contact › Email – me@petermccormack.com › Instagram – /mccormack555 › X/Twitter – https://x.com/petermccormack/

CONNECT WITH CONNOR LEAHY › X/Twitter – https://x.com/NPCollapse › Website – https://www.ettf.land/

SPONSORS › IREN - https://www.iren.com/ › Ledger - https://www.ledger.com/

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

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.

Transcript

Click on a timestamp to play from that location

0:00.0	So it's very important to understand is that we do not understand intelligence. We don't know how the brain works, you know, we have a bunch of guesses, but we sure as hell don't know how it works, and we sure as hell don't know how these neural networks work either. So we've built something, but we don't understand how it was? That's exactly correct. So it's kind of like magic? Yes, absolutely. It's kind of like looking into a petri dish.
0:21.6	We do not know what our AIs can do until we make them.
0:23.6	And even after we make them, like we don't know, like we don't know what chat GP-6
0:28.6	can do until it's done.
0:30.6	None of the engineers at AI know what it will be able to do until it's done.
0:35.6	And this is very, very different from other forms of engineering.
0:39.4	So even if the a arrow doesn't kill us all, it can still dethrone us as a species. What is our
0:47.6	role? What is our purpose? Where do we exist? Exactly. I think this happens before extinction
0:51.7	happens. Like the thing I expect to happen is that one day we wake up and we're just not in control anymore.
0:57.5	And I don't think we'd all fall over dead or anything like that.
1:00.0	I don't think extinction happens right away.
1:01.4	But we won't be in charge.
1:03.1	We won't be in control.
1:05.7	This show is brought to you by my lead sponsor, Iron, the AI cloud for the next big thing.
1:11.3	Iron builds and operates next generation data centers and delivers cutting edge GPU infrastructure,
1:16.7	all powered by renewable energy.
1:18.8	Now, if you need access to scalable GPU clusters or are simply curious about who is powering
1:24.2	the future of AI, check out iron.com to learn more, which is iri-en.com.
1:30.5	Good morning, Connor. How are you? I'm doing great today.
1:33.2	Really good to meet you. So I'm very intrigued to talk to you because your background is actually
1:38.5	working in the engine historically of building LLMs. And we've been talking about AI in the show quite a bit.
1:45.6	But I'm conscious, I know nothing about the engines.
	...

Please login to see the full transcript.

Previous episode | Next episode

Disclaimer: The podcast and artwork embedded on this page are from Peter McCormack, and are the property of its owner and not affiliated with or endorsed by Tapesearch.

Generated transcripts are the property of Peter McCormack and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.