Biden meets with tech leaders to discuss future and regulation of artificial intelligence

Read the Full Transcript

• Geoff Bennett:

  In San Francisco today, President Biden convened a meeting of artificial intelligence experts to weigh its risks and opportunities and consider the role of the federal government in regulating the technology.

  Following all of this closely is Ryan Calo. He's a professor of law and information science at the University of Washington.

  Welcome back to the program.

  And, look, given all of the warnings about A.I.'s present and potential future, concerns about discrimination, misinformation, privacy, how should the federal government approach the notion of regulating A.I. without stifling innovation?

  Ryan Calo, University of Washington: Well, it's a hard question.

  I mean, I think, on the one hand, the Biden administration recognizes the transformative power of artificial intelligence and is worried about rivalrous nations like China getting better at A.I. than we are.

  At the same time, it's hard to deny the fact that there's some real consequences to the deployment of this technology, especially in the state that it's in, where it's not totally ready for prime time. And those things are things like you just mentioned, privacy, misinformation, bias. And that's what the experts are talking about today.

• Geoff Bennett:

  The European Union, as you well know, is taking steps toward passing what would be the first major law regulating artificial intelligence that would ban real-time facial recognition. It would require more disclosure for programs that use artificial intelligence.

  How is the E.U. moving so quickly? And could be — could what they're developing be a model for other countries?

• Ryan Calo:

  Absolutely.

  Just recently, the European Parliament passed a draft of the A.I. Act that does those things that you mentioned. It heavily regulates some risky A.I., bans other stuff that they're really not interested in seeing, and then has a real light touch with a lot of other stuff.

  And what the E.U. experience shows is that it is possible and, indeed, probably wise to have some regulation around artificial intelligence. Why are they way out ahead of us? Because there's just a different culture, a different thought process about technology in the E.U.

  I think, in the United States, we're much more concerned about making sure that we capture those — the productivity and the efficiencies and the innovation and even also the military edge that something like A.I. could confer, and very concerned about killing the goose that lays the golden egg, whereas I think that, in Europe, they're much more precautionary.

  And so what they're doing could be a role model for us.

• Geoff Bennett:

  How long might that process take in this country? As I sit here and talk to you, it strikes me it took, what, 50 years to regulate cigarettes. Regulating social media is still an open question.

  Might this come down to the companies themselves having to establish their own guardrails? Right now in Silicon Valley, A.I. is anything anybody can talk about.

• Ryan Calo:

  I hope not.

  I mean, so all we have right now really is this blueprint for an A.I. bill of rights, which is awfully similar to a similar blueprint that the Obama administration worked on many years ago. In fact, I worked with that administration that bill of rights. We haven't moved a lot.

  So what would have to happen, of course, it is that Congress would have to get involved or individual states would have to pass laws. And, again, Congress and individual states are going to be very worried about regulating too heavily.

  But it shouldn't really be left to the companies to make these kinds of decisions, because, left to their own devices, as we can see with past technologies like the Internet, they're not going to fully address or mitigate the big range of harms, whether to the environment or job displacement, bias, privacy, misinformation.

  They're not going to internalize those costs unless the government, I think, makes them do it. So I'm hopeful that we will we will make a move like Europe, but I'm not holding my breath.

• Geoff Bennett:

  In the minute or so that we have left, can we talk about the promise of A.I., setting aside the risk assessment for a second?

  Because what generative artificial intelligence can do is really extraordinary, writes original essays, and can compose music, can generate code. What do you see on the horizon?

• Ryan Calo:

  Well, I mean, the fact that we have moved from artificial intelligence spotting patterns to actually coming up with output that is as good as a human in some contexts is really powerful.

  And so my hope is that it will unleash a lot of productivity and creativity within the arts, within music. I also hope it will help people with differences, people who are differently abled maybe accomplish some of their goals. I know that my own son made a birthday card using artificial intelligence who struggles with reducing things to writing.

  So there is a lot of potential there, and we shouldn't — we shouldn't deny that. But there are also some real harms. And if this is truly a transformative technology of our age, one of the things that needs to transform is our law and our legal institutions.

• Geoff Bennett:

  Ryan Calo is professor of law and information science at the University of Washington.

  Thanks again for being with us.

• Ryan Calo:

  Thank you.