Lawfare Daily: Peter Salib on AI Self-Improvement

The Lawfare Podcast

The Lawfare Institute

News, International Law, Foreign Policy, Military, Intelligence, Government, National Security, Rule Of Law, Constitutional Law, Law, Terrorism, Current Events, History, Diplomacy, Politics, International Relations

4.7 • 6.4K Ratings

🗓️ 20 May 2024

⏱️ 48 minutes

🔗️ Recording | iTunes | RSS

🧾️ Download transcript

Summary

In foundational accounts of AI risk, the prospect of AI self-improvement looms large. The idea is simple. For any capable, goal-seeking system, the system’s goal will be more readily achieved if the system first makes itself even more capable. Having become somewhat more capable, the system will be able to improve itself again. And so on, possibly generating a rapid explosion of AI capabilities, resulting in systems that humans cannot hope to control.

Alan Rozenshtein, Associate Professor of Law at the University of Minnesota and Senior Editor at Lawfare, spoke with Peter Salib, who is less worried about this danger than many. Salib is an Assistant Professor of Law at the University of Houston Law Center and co-Director of the Center for Law & AI Risk. He just published a new white paper in Lawfare's ongoing Digital Social Contract paper series arguing that the same reason that it's difficult for humans to align AI systems is why AI systems themselves will hesitate to self-improve.

To receive ad-free podcasts, become a Lawfare Material Supporter at www.patreon.com/lawfare. You can also support Lawfare by making a one-time donation at https://givebutter.com/c/trumptrials.

Support this show http://supporter.acast.com/lawfare.

Hosted on Acast. See acast.com/privacy for more information.

Transcript

Click on a timestamp to play from that location

0:00.0	The following podcast contains advertising.
0:04.0	To access an ad-free version of the Lawfair Podcast,
0:08.0	become a material supporter of Lawfair at Patreon.com slash Lawfair. That's Patreon.com
0:16.4	slash Lawfair. Also check out Lawfair's other podcast offerings, rational security, chatter, lawfare no bull, and the aftermath.
0:28.6	Did you know that it takes just 30 seconds to get a quote from We Buy Any Car?
0:36.0	And it took me just three seconds to tell you that.
0:38.0	So, uh, what I'm meant to do now? Any ideas?
0:42.0	Oh, I know.
0:43.0	Oh! I know!
0:44.0	H.
0:45.0	H. Just sold my car.
0:49.0	To find out how much your car's worth in 30 seconds, enter your
0:54.2	reg number now at we buy any car.com.
0:56.3	Admin fee may apply. For more information, visit we buy any car
1:01.0	dot com slash info. Visit We Buy Any Car.com. In fact, when models are acting in the world autonomously on their own,
1:11.1	I think it's worth asking whether we are certain that they would not, for example,
1:16.2	take advantage of their programming ability to, you know, access secure systems
1:21.9	and disrupt the competitors of the companies for whom they are
1:26.4	acting.
1:27.4	It's the Law Fair Podcast.
1:29.2	I'm Alan Rosenstein, associate professor at the University of Minnesota Law School and senior editor at law fair
1:34.2	and I'm here today with Peter Salib
	...

Please login to see the full transcript.

Previous episode | Next episode

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

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