BENSON BYTE: Eugene Volokh - "The Medical Establishment Has Squandered Much Of Its Credibility As A Result Of COVID"

SPEAKER_1 (0:00)

Now it's time for another Guy Benson Show, Benson Bite, where we bring you the day's most interesting interviews in a bite-sized package. For the full episode, visit guybensonshow.com or wherever you listen to podcasts.

Guy Benson (0:17)

Joining us now here in studio at the Hoover Institution is Eugene Volokh. He is a senior fellow at Hoover, also has been for years a professor of law at UCLA. It is very good to see you, welcome.

Eugene Volokh (0:30)

A great pleasure to be here. Thanks for having me.

Guy Benson (0:32)

I want to start with just some legal issues, some First Amendment related issues. I know this is your bread and butter. And I was actually daydreaming about this the other day, and you're the perfect guest to ask, in the age of AI and so much content being generated by AI, at the direction of real people, sometimes not, perhaps in the future, there'll be less of a human hand in this sort of thing.

How does that content, AI created, interact with the First Amendment and certain free speech protections?

Eugene Volokh (1:05)

Yeah. So my thinking, and courts have not really fully reached that. There's been a little bit of discussion, but very little. But my thinking is that the government should have the same power to regulate AI output, that is, say, generative AI, text, images, video, that it has for ordinary speech, which is to say, it's sharply limited. But there are exceptions to First Amendment protection. The reason I think that that's so is because certainly today, at least, when you see AI output, it was indeed generated by a cooperative process of humans prompting the AI and then often reprompting and modifying it in various ways. And the AI. AI is a tool. It's like a camera, let's say.

The government can't say, well, we're going to ban you from using cameras to express certain viewpoints.

Or if there's some way of making sure camera manufacturers not allow photographs of, let's say, police conduct and the like, that would be unconstitutional. Not because it somehow interferes with the right of AI. AI's don't have rights. Or even of the AI companies or the camera manufacturers. It's because it interferes with the rights of people, the people who would be using the cameras or the AI to speak, to generate their own speech, to gather information, to learn things and the like.

Guy Benson (2:36)

What could be a borderline or thorny case? Because my imagination maybe isn't good enough to think through, and often justices in oral arguments will have a bunch of hypotheticals. What if this and they'll spar with the lawyers? Where would that line start to get blurry?

Eugene Volokh (2:53)

Well, so let me give you two examples of things where there might be permissible restrictions, not because somehow AI is special, but because First Amendment rights are broad, but not absolute. So one example, there have already been nine cases in American courts where people sue AI companies for outputting things that defame them.

False, hallucinated factual assertions that damage the reputation. Some of these cases are still pending, and some courts have allowed these cases to go forward.

Guy Benson (3:22)

Like so-and-so is convicted of a crime, it just was not true.

Eugene Volokh (3:25)

Exactly right. Exactly right. And we know the New York Times has First Amendment protection, doesn't mean it can publish libel, knowing libel, let's say.

Likewise with AI output. Now, there are interesting questions. It's really important in libel cases to determine in some cases where whether the speaker spoke with so-called actual malice, a kind of deceptive term which means knowing or reckless falsehood. In other cases, you have to ask whether the speaker acted negligently. Well, what does knowledge and recklessness and negligence mean for AI output?

SPEAKER_4 (3:58)

Interesting questions.

Eugene Volokh (3:59)

I think it probably turns mostly on what the AI company, what the human beings of the AI company knew. But that's the kind of thing that courts have to deal with and are already dealing with.

For AI, just as they would be dealing with for other speech. Another example is there are lawsuits where the claim is that AI companies or products were designed in a way that would promote very emotionally engaging conversations, and sometimes they were engaged in with people who then as a result, committed suicide or attempted murder or something along those lines. These are interesting questions.

They held liable. Right. There are exceptions to First Amendment for solicitation of illegal conduct. What if the AI output said, Oh, you should kill so and so? It's an interesting question. How do we apply solicitation logic there? Some of these cases involve minors. Now, there is a First Amendment precedent that says that minors do have very broad free speech rights for pretty much everything other than sexually themed speech, including the case involved violent video games. And it struck down a law that banned minors access or that limited minors access to violent video games.