"So one way is to put in a process-orientated approach making the federal government and companies think about the AI they're deploying and how they themselves make sure that the AI they're using doesn't harm people," Rep. Lieu explains. "Another... is to make sure that our existing legal structure can go after harms caused by AI, whether it's discrimination in hiring or giving out of loans."
AKIKO FUJITA: Well, lawmakers in the US are pushing forward in their effort to regulate artificial intelligence. A group of nonpartisan congressmen unveiling a bill that would require federal agencies to abide by AI guidelines put forth by the Commerce Department last year. One of those behind the proposed bill, California representative Ted Lieu joins us now to discuss.
Congressman, it's good to talk to you today. Let's talk about what's at stake first. What is the risk you are looking to address through this bill?
- Thank you for your question. Let me first say that the S&P 500 closed at a record high last Friday, the Dow exceeded 38,000 for the first time in US history yesterday. So the economic policies of the Biden-Harris administration and Congress are working. And part of those policies includes making sure that artificial intelligence continues to innovate, but does so in a safe way that doesn't harm people.
AKIKO FUJITA: So to follow up though specifically, Congressman, when you talk about this bill, what is the risk you are looking to address?
TED LIEU: And so what this bill does is it mandates that the standards set forth by the National Institute of Standards and Technology be applied to both the federal government, as well as any companies that want to contract with the federal government. This is known as an AI risk management framework. It's already in existence in both the private sector and public sector. And what it does is it makes companies and institutions think about how they're using AI in their own corporations or their own organizations to make sure that it's safe, that there's AI governance that doesn't unnecessarily harm people, and it doesn't have effects that will really continue to harm either their company or their organization.
RACHELLE AKUFFO: And Congressman, as we look at the comparisons to some of the EU legislation, theirs is more done on an impact assessment. How is it different, the legislation that you're putting through? How are you assessing how you're categorizing these AI risks?
TED LIEU: So our bill is more process orientated. It makes these companies who want to contract with the federal government, as well as the federal government itself, go through an entire process of thinking about AI, how they deploy it, how they test to see that it doesn't cause harm, how they go back and then try to fix it if, in fact, it is causing harm. So it is more process orientated. I also note that this bill is not only bipartisan, it's also bicameral with senators Moran and Warner having introduced it in the US Senate as well.
AKIKO FUJITA: And what is your expectation in terms of the timeline for passage or the potential for it?
TED LIEU: So we, unfortunately, do have a house Republican caucus that is rather chaotic. Last year, we were just trying to prevent stupid stuff like government shutdowns and default on our nation's debt. Now that we've managed to keep our lights on, hopefully, we're going to be able to work on other issues such as AI regulation. And I'm hopeful that this bill can be passed sometime this year.
RACHELLE AKUFFO: And Congressman, what does that mean in terms of trying to coordinate? Obviously, something like AI is essentially borderless. How do you coordinate that with other countries who are also taking sort of a piecemeal approach and have a very different set of rules depending on where you are?
TED LIEU: That is a great question. Unfortunately, there's very little coordination right now. The UN is going to put together a summit later this year that's going to bring together all the countries of the world to have more coordination on artificial intelligence. Another thing we're going to see is these jurisdictions try different approaches to regulating AI.
One of the things that lawmakers do is we plagiarize and so we're going to look at how this AI regulation goes in Europe. If it goes well, we're going to copy it. If it's a disaster, we're not going to copy it. So it allows us a chance to also see how other jurisdictions try to regulate AI.
AKIKO FUJITA: Artificial intelligence, just the very latest in terms of tech regulation that lawmakers have had to oversee, the argument is often you've got to find that balance between regulating the risks of a technology, but doing so without stifling innovation, especially in a place like AI where there is really a global race to be the leader. How do you find that balance?
TED LIEU: So one way is to put in a process-orientated approach making the federal government and companies think about the AI they're deploying and how they themselves make sure that the AI they're using doesn't harm people. Another, as mentioned earlier on your show, is to make sure that our existing legal structure can go after harms caused by AI, whether it's discrimination in hiring or in the giving out of loans. And then if those things don't work, then we're going to have to look at a firmer regulation to simply say, you can't do certain things with AI or if you cause certain harms, we're going to make sure you can get sued.
RACHELLE AKUFFO: And we know this isn't an issue going away anytime soon. We'll continue to track it. We appreciate you taking the time to join us this morning. Congressman Ted Lieu, thank you so much.