Bridget Dooling and Mark Febrizio on Robotic Rulemaking
The Lawfare Podcast
The Lawfare Institute
4.7 • 6.4K Ratings
🗓️ 4 May 2023
⏱️ 36 minutes
🧾️ Download transcript
Summary
At the core of the regulatory state is the notice and comment process. Agencies propose what they're going to do, the public gets to comment, and agencies have to respond to those comments. It's an imperfect system, to be sure, but it's fundamental to making sure that agencies act with good information and with democratic legitimacy.
So what happens when those comments start being made not by people, but by ChatGPT or other large language models? Or how about when agencies themselves use these AI tools to analyze the comments they receive, or even perhaps to write the regulations themselves?
To talk through these issues, Alan Rozenshtein, Associate Professor of Law at the University of Minnesota and Senior Editor at Lawfare, spoke with Bridget Dooling and Mark Febrizio, both of the George Washington University Regulatory Studies Center. They spoke about their recent Brooking Institution report on the issue and how they think the regulatory state should deal with generative AI.
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 law fair. |
| 0:14.0 | That's patreon.com slash law fair. |
| 0:18.0 | Also, check out LawFair's other podcast offerings, |
| 0:22.0 | rational security, chatter, law fair no bull, and the aftermath. |
| 0:29.0 | The LLMs can help generate potentially millions of comments |
| 0:37.0 | that are sufficiently different from each other. |
| 0:41.0 | If you put yourself in the shoes of the agency staff reading those comments, |
| 0:45.0 | it might not be clear that 85% of this comment is identical to 10,000 others that we've received, |
| 0:52.0 | because there might be enough differentiation between the comments that the LLM generates, |
| 0:56.0 | that those tools might have to catch up with the tools that are available to the commenters. |
| 1:02.0 | Right now, we're not aware of tech inside the government that helps them distinguish |
| 1:08.0 | tremendously different comments in this way. |
| 1:12.0 | That's akin to the thing that they have already, which allows them to say, |
| 1:16.0 | this paragraph is 85% the same as the 10,000 other comments that we got. |
| 1:22.0 | I'm Alan Rosenstein, Associate Professor of Law at the University of Minnesota |
| 1:26.0 | and Senior Editor at LawFair, and this is the LawFair podcast for May 4, 2023. |
| 1:32.0 | At the core of the regulatory state is the notice and comment process. |
| 1:36.0 | Agencies propose what they're going to do, the public gets to comment, |
| 1:40.0 | and then agencies have to respond to those comments. |
| 1:42.0 | It's an imperfect system to be sure, but it's fundamental to making sure that agencies act with good information |
... |
Please login to see the full transcript.
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.
Copyright © Tapesearch 2026.