Lawfare Daily: Josh Batson on Understanding How and Why AI Works

The Lawfare Podcast

The Lawfare Institute

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

4.7 • 6.2K Ratings

🗓️ 30 May 2025

⏱️ 43 minutes

🔗️ Recording | iTunes | RSS

🧾️ Download transcript

Summary

Josh Batson, a research scientist at Anthropic, joins Kevin Frazier, AI Innovation and Law Fellow at the Texas Law and Senior Editor at Lawfare, to break down two research papers—“Mapping the Mind of a Large Language Model” and “Tracing the thoughts of a large language model”—that uncovered some important insights about how advanced generative AI models work. The two discuss those findings as well as the broader significance of interpretability and explainability research.

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/lawfare-institute.

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.
0:04.4	To access an ad-free version of the Lawfare podcast, become a material supporter of Lawfare at
0:11.5	patreon.com slash lawfare. That's patreon.com slash lawfare.
0:18.2	Also, check out Lawfare's other podcast offerings, Rational Security, Chatter,
0:25.2	Lawfare No Bull, and The Aftermath.
0:31.8	Ever wish your brain had a psychic that could help fast forward you through this long afternoon
0:37.1	in the blink of an eye.
0:39.5	Whilst we can't promise that, Baroque and Mind contains vitamin B5 to support mental performance,
0:45.1	iron to support energy release and is enriched with Spanish sage extract. Find it now at major retailers
0:50.8	in two zingy flavours, citrus and berry. Just drop fizz and enjoy. Your mind's the
0:56.6	hero. We're just the psychic. Baroque mind. And I think this is one of the things I love
1:08.0	about interpretability is because these models aren't programmed, we don't know what strategies they've learned. And this is the case where it had sort of learned,
1:15.7	at least in this poetry context, to plan ahead to write something good, even though we never
1:20.7	told it to do that. It's the Lawfare podcast. I'm Kevin Frazier, the AI Innovation and Law
1:27.3	Fellow at Texas Law and a senior editor at Lawfare,. I'm Kevin Frazier, the AI Innovation and Law Fellow at Texas Law and a senior
1:29.3	editor at Lawfare, joined by Josh Batson, a research scientist at Anthropic.
1:35.6	You know, I would like every judge to have to produce rulings that seem at least as reasonable
1:42.0	to another party as that produced by the system of AIs.
1:46.0	Right? You know, I think that like we should be holding humans to at least the standard of what the AIs could do in terms of the quality of their, like, written opinions.
1:56.0	Today we're talking about Anthropics' leading work to make the, quote, black box, end quote, of AI a little less opaque.
2:04.8	This line of research has significant ramifications as AI becomes a part of ever more sensitive processes, such as hiring, admissions decisions, and medical diagnoses.
2:15.6	Anyone who has loosely heard about AI,
	...

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.