WATCH: Can AI companies be trusted to regulate themselves?

As artificial intelligence races into nearly every corner of American life, a fierce debate is unfolding over how much the government should step in to control it. Should AI companies be held legally responsible when their products cause harm? And should protecting children from AI’s risks outweigh concerns about stifling innovation? 

Daniel Castro, President of the Information Technology and Innovation Foundation, and J.B. Branch, Director of Federal AI Governance and Technology Policy at Public Citizen, join Billy Hallowell on “The Dialogue” to debate these urgent questions, clashing over corporate liability, child safety safeguards, and whether Washington can regulate AI without falling behind China.

The claims discussed are those made in the debate and do not represent the views of The Washington Times.

[HALLOWELL] Welcome to “The Dialogue” from The Washington Times. In this episode, we’re covering an important question: Should the U.S. government impose stricter regulations on artificial intelligence? Now, with AI becoming such a massive part of Americans’ lives, this conversation is only going to heat up. I’m joined by Daniel Castro, president of the Information Technology and Innovation Foundation and J.B. Branch, director of federal AI governance and technology policy at Public Citizen.

We start off each show with something called “The Pitch.” And this gives you each 30 seconds to tell us where you stand on a central statement surrounding our topic. The statement today is this: The government should impose stricter regulation on artificial intelligence.

[CASTRO] I’ll say I partially agree. I mean, I think there’s a lot of regulations that already apply to AI. What I’m concerned, though, is that we’re not moving fast enough in terms of adoption. And many of the proposals I’ve seen for putting heavier regulation on AI looks at the technology and treats it all the same. It treats AI being used in healthcare the same as AI being used to schedule a vacation for someone. And I think that’s the problem.

We don’t want to have too many regulations that hold back productive uses of the technology. We want targeted rules that address very specific harms and situations.

[HALLOWELL] J.B., same statement: The government should impose stricter regulation on artificial intelligence. Where do you stand?

[BRANCH] I definitely think we need to have stricter regulations on artificial intelligence. I mean, we’re basically seeing a bit of a replica of what happened in the early 2000s, where tech companies were basically saying we can regulate and police ourselves. We’re the experts. And what ended up resulting was roughly 20, 25 years of harms, myths and disinformation, harms to minors, addicted to designs.

Today, AI companies are largely recycling that same playbook. They’re rushing to market with their products without concerns for harms. And they’ve learned from the last sort of tech revolution that they can do that and maybe settle for pennies on the dollar for the harms that were caused. At Public Citizen, we think that America can lead AI just without Silicon Valley basically writing their own rules.

[HALLOWELL] Those are two perspectives. And I think this is interesting because you can have multiple perspectives almost on this statement, right? Because it’s broad and there’s a lot that goes on within it. So we want to dig deeper. 

This is something that we call our Focus section. And we present three to four claims and we have our guests then debate each claim. And they’re based on the broader topic. And, you know, this is, I think, an important one. And I’m going to throw the first claim out. I’m going to let you guys dive in. J.B., I’m going to start with you first. 

Here’s the claim: AI companies should be held legally responsible when their technology produces harmful or dangerous outcomes.

[BRANCH] Yeah, I mean, to me, that’s a very sensible statement.

Basically, what’s being outlined here is: should AI companies be required to follow the law? And I’m a hard yes on that. These tech companies have really perfected a business model that really focuses on maximizing profits, capturing market as much as possible, and releasing products to market to capture that market as quickly as possible. AI companies similarly are following a similar playbook. And if they’re not being held responsible through litigation or state attorneys general or other mechanisms, you give them an inch, they’ll take a mile, they’ll take two, 10 miles. Some folks get concerned that this litigation would slow down innovation. I think a great example of this was the Meta litigation that happened from New Mexico and California. Meta ultimately ended up settling for roughly $400 million. But in that same amount of time, they accumulated $55 billion of revenue. $400 million basically represents one day’s worth of work for Meta. And so again, even when they are being held accountable, it’s a fraction of the total cost of money that they’re making.

[HALLOWELL] Daniel, is there anything in that that you want to respond to?

[CASTRO] Yeah, a lot. I mean, first of all, there’s nothing wrong with companies pursuing profits. I mean, we’re talking about what’s driving the American economy right now. We want these companies to be successful. We want them to grow. And we want them to pursue success.

I don’t think we should be making this comparison to social media. Social media services are very specific. They’re used in a specific way. AI is a general-purpose technology. Now, the question about should they be liable for their uses, I mean it really depends on the specifics here. If we’re talking about an AI company that’s providing a specific service that has a harmful outcome that they could prevent or that they knowingly could stop, of course, they should be liable. And we have existing tort law that allows for that. 

I think the concern is what we’re seeing there are proposals that would make developers of AI technology responsible for all the various downstream uses, two and three levels down that people could have. That’s like saying a company that makes a screw is going to be responsible for how someone else decides to melt that screw down into something else and use it in another product. You get so separated, you start losing sight of what these companies are trying to achieve. And the risk is you have regulations that actually hold back the development of better AI because companies are scrambling to basically mitigate the risk of potential lawsuits instead of trying to mitigate actual harms from their technology by designing products that are safer and better.

Watch the video for the full conversation.