“I’ve been thinking about this David Foster Wallace quote more and more recently: The technology is just gonna get better and better and better and better. And it’s gonna get easier and easier, and more and more convenient, and more and more pleasurable to be alone with images on a screen, given to us by people who do not love us but want our money. Which is all right. In low doses, right? But if that’s the basic main staple of your diet, you’re gonna die. In a meaningful way, you’re going to die.” —Eugenia Kuyda, the founder of Replika, one of the first companies built around offering A.I. companionship. The cognitive dissonance, she said, is “very high.”

And now, here’s Leigh Ann’s conversation about A.I. with Senator Rounds…

Anyone who’s ever watched a congressional hearing focused on tech policy—say, social media, or digital data collection—knows that our legislative body doesn’t have the most sophisticated understanding of any current technology, let alone the rapidly advancing field of artificial intelligence. That’s why then-Senate Majority Leader Chuck Schumer tapped Sen. Mike Rounds, the affable A.I. enthusiast and South Dakota Republican, during the last Congress to help educate his colleagues on the issue—including by convening top tech leaders such as Elon Musk, Mark Zuckerberg, and Sam Altman to give a borderline geriatric Senate a crash course in Artificial Intelligence 101.

Since then, of course, the technology has found an unlikely champion in Donald Trump, a man who famously has avoided using email, and yet returned to office pursuing a win-at-all-costs agenda for global A.I. supremacy—including a designated A.I. czar and releasing an “action plan.” Rounds isn’t far behind in his zeal, and is working the other side of Pennsylvania Avenue to push for more rapid A.I. adoption in finance, health, science, and more.

And yet, when I interviewed Rounds at the Puck Power Breakfast at the Riggs Hotel this morning, he notably disagreed with the administration in several respects. He thinks Congress—and the states—need to regulate A.I., as well as support wind and solar sources to help meet its surging energy demands.

Leigh Ann Caldwell: How do you view regulations and A.I, and what do you think Congress should do? Should they stay out of it or get involved?

Mike Rounds: If Congress is ineffective, then states will take it into their own hands. Congress is ineffective in a lot of cases, and because of that, states look at this and say, We might not be able to protect the rest of the country, but we can sure protect our people. My opinion is, Congress needs to take a leading role laying out standards for everybody in the country. But we have to respect the fact that if we’re not going to do it, [the states are] going to do it. We need to keep our eyes on what [the states] are doing, but we need to step forward as well.

What does that look like?

You’ve got different committees that all have different points of view, but you’ve got to bring people together. You have a Justice Department that’s going to have one point of view based on what the existing laws are, and a Judiciary Committee that’s going to take a point of view. Commerce [is] going to be listening to the different business interests and talking about how they should regulate them and create new law. You’ve also got the rest of Congress out here saying, We’re getting feedback from people in lots of different states about privacy concerns, about identifying what’s real and what’s not. Those are the types of things where I think Congress is going to have to get more actively involved—but it’s just the tip of the iceberg.

So Congress is behind?

Definitely. The last time we actually did an appropriations bill was probably December 2023. How the hell are you supposed to make changes when you’re doing continuing resolutions and you don’t pass bills? You’ve got to have a working, operational, functional decisionmaking body. And it’s why I come back and tell folks we have a lot of work to do up here.

Three years ago, I got a briefing that was at a Special Access level, and it dealt with what the military had to do to integrate artificial intelligence into the defense of our country. As I listened, all I kept thinking was, My gosh, this is stuff we got to get started on right now. It’s been three years, and we haven’t had an appropriations process in place since then to actually put the money into places so that we could develop the A.I. that will impact people’s lives forever. We’re three years behind from where we should be, just because we’re not doing appropriations.

Absent congressional regulation, should companies who are developing this [technology] put in their own safeguards?

For them, it’s better to have the safeguards built in rather than be liable for something going south on them. For those that don’t, they’re probably not going to be around very long, because the cost of fixing the problems could be significant. On the other hand, you’re going to have companies out there saying, We got some real whack jobs out there that aren’t taking care of their stuff. They're moving forward and breaking things. I got a system that will help fix it, and I can make a lot of money doing that. We can’t stop the innovation that’s going on, but we can help it by laying out the framework in which we’re going to develop artificial intelligence, and how it’s going to be applied in the future.

The limiting factor is energy. We have about 1,300 gigawatts of energy right now that we use in the United States. Between now and 2028, we’re going to need between 75 and 100 gigawatts of additional power just for A.I. development in the United States. China is developing a third of what we’re doing right now, every single year; so they’re doing 400 gigawatts a year in new development. Our supply chains can’t compete with that today.

The administration doesn’t believe in an all-of-the-above strategy, shutting down wind and solar projects around the country. Is that a mistake?

If they’re actually shutting them down, the answer is yes, because we need all of the above. We can’t build without turbines until we literally go to space-based power—which, by the way, is not that far away—where you actually capture energy in space. Eventually, we will produce, some people say, anywhere between 30 to 50 percent of all the energy needs of our country in space.

How far away is that?

China is doing it now; China has been working on it for three years. Let’s just say it hasn’t gone unnoticed by our folks. You’ve got commercial interests that are looking at different ways of doing it today. Easily within the next five years, you’re going to have space-based energy collection systems.

Is it necessary that the U.S. takes a posture that China is an adversary in A.I.?

Yes, no question. I would love to say that China is a competitor and not an adversary, but everything we see says that they’re an adversary. There’s a program right now where they’ve decided that their part of the world belongs to them, and they believe that it should be Chinese standards, and it should be Chinese dominance in the Pacific Rim region. They’re moving in that direction—nothing that we’re seeing in the intel community suggests anything else.

They are trying to put themselves in a position that when they do decide to make advancements in the Pacific Rim region, that we will look at this and say, Maybe we really don’t want to challenge them. Cybersecurity is where it will start. No matter what happens out there, the early warning for us will be when they start to try to take out our cybersecurity systems in the United States and [those of] our allies. That will be the first warning sign that things are beginning to happen on a kinetic basis elsewhere in the world.

Read more from the interview here.

Following a very public bulking phase, wherein Mark Zuckerberg handed out eye-popping employment offers to A.I. engineers across the ecosystem, Meta has now entered a cutting phase: It’s laying off 600 people within its A.I. division. [Axios]

Reddit has filed a lawsuit against Perplexity and a few others for training its A.I. “answer engine” on Reddit content without permission (or, more importantly, compensation). [Adweek]

And even as chatbots and language models are positioned as a replacement for traditional search engines, a recent, large-scale study identified a potential impending bump in the road: Such tools misrepresent news content around 45 percent of the time. So… read Puck, not ChatGPT. [EBU]

That’s all for today. I’ll see you next week.

Ian

Need help? Review our FAQ page or contact us for assistance. For brand partnerships, email ads@puck.news.

You received this email because you signed up to receive emails from Puck, or as part of your Puck account associated with {{customer.email}}. To stop receiving this newsletter and/or manage all your email preferences, click here.

Puck is published by Heat Media LLC. 107 Greenwich St, New York, NY 10006