JUANA SUMMERS, HOST:
Conversations about the potential dangers of AI hit a fever pitch this week in Congress. Then the House skipped town on Wednesday and headed back home to the campaign trail. That means, unless they change the schedule, nothing will happen on AI regulation until at least after the midterm election. So where does that leave us? NPR congressional reporter Eric McDaniel is here to help us understand. Hi there.
ERIC MCDANIEL, BYLINE: Hi, Juana.
SUMMERS: So, Eric, I mean, this week, there has been just, frankly, some seriously scary talk about the power of AI. Why didn't the House act on AI regulation before leaving town?
MCDANIEL: I mean, but by the time all of that was coming down, we knew they just had three or four days left. And in part, Republicans decided they needed to get home ASAP PDQ in order to campaign. They're at real risk of losing control of the House. Also, there were some thorny votes they seemed to really not want to take, including an impeachment effort targeting Secretary of Defense Pete Hegseth that was introduced by Kentucky Republican Thomas Massie, a perpetual thorn in the side of the administration.
But it also isn't like Congress being slow is an AI-specific issue. We've reached a kind of partisan maximalism where none of the incentives are aligned for cooperation. They haven't even passed regular appropriations bills to fund the federal government in anything close to a regular manner since, like, the 2019 fiscal year.
SUMMERS: Do you think that part of this is that lawmakers, like, well, frankly, a lot of us don't understand enough about the technology to make the rules?
MCDANIEL: Generally, I'd say people don't think of Congress' job as needing to understand things at the highest real expert level, right? Their job is to set up the infrastructure for experts to act, to decide what the goalposts are, say that models should not produce outcomes that are dangerous for children in these ways, or models can't be used to do this kind of fraud. You don't hear anyone making the case, for instance, that we can't regulate nuclear bombs without Speaker of the House Mike Johnson knowing how to build a bomb or Chuck Schumer needing to understand pharmacology in order to regulate prescription medicines. Lawmakers need to know what smart policymaking looks like, right? They need to know which academics and industry experts they should listen to and trust and go from there. That's what lawmaking is.
SUMMERS: OK, so if they can't pass anything this year, are they just going to be starting from scratch with a new Congress next year?
MCDANIEL: No, there are so many ideas knocking around. In fact, there was a bipartisan task force in the last Congress. So two years ago, there's a bipartisan framework that was proposed to this Congress. There's a domestic - democratic task force, sorry, that the House is working on. So, so many ideas floating around about the substance of regulation. But one big thing they have to figure out first is what the shape of the rules look like.
SUMMERS: Right. Because, I mean, there are these competing philosophies at play about how to think about regulation. So, I mean, what does that debate even look like right now?
MCDANIEL: Well, for one, who's the boss, right? Is it the federal government or the states? Should we have federal rules and ban states from making their own? Should we leave it exclusively up to the states? Should states be able to go further than federal laws? So that's the start. And the second thing is, what does a regulator look like? Is it an outside group of AI companies, like the Motion Picture Association does for the film industry that comes up with voluntary standards? Is it a government body, like the Securities and Exchange Commission, or is it something that's kind of half in, half out, more like FINRA, for all the finance nerds out there that can refer violations to the government?
SUMMERS: Last thing, Eric, how is all of this playing out on the campaign trail?
MCDANIEL: You know, when I talk to candidates, I ask a kind of stock question, like, is there anything that I forgot to bring up that you think is important to talk about? And they've all hopped right to data centers. There are these kind of amazing literary objects, right? Because they're real structures in real communities with real trade-offs. But they're also this vessel to hold the worries that I imagine many of us have about AI. Is it going to present a new age of safety threats? Is it going to mean my job is no longer meaningful, that there's no place for my kids in the workforce? Am I going to get smooshed by a self driving car? So in the same way that we saw the offshoring of American manufacturing jobs drive an anti-immigrant political shift, I imagine we're about to find out together where society places its frustrations with AI and the economic revolution that it brings.
SUMMERS: NPR's Eric McDaniel. Thanks as always.
MCDANIEL: Thank you. Transcript provided by NPR, Copyright NPR.
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.
