B (2:46)

And in order to make a huge amount of money from these technologies, you need huge, lucrative applications. How are we going to make enough revenue to justify hundreds of billions of dollars of capital expenditures that's required to train these models?

A (3:03)

Consider this on Wall street and in Silicon Valley. Even people who fully believe in the potential of AI are now warning that right now the AI market may be in a bubble of overinflated expectations. What if we are in a bubble of and what if that bubble bursts? From npr, I'm Ilsa Chang. Rhinoplasty, the humble nose job. It's one of the most common facial plastic surgeries performed today, but it's been around for over 2,000 years. If you're wondering why on earth would doctors thousands of years ago need to reconstruct noses, that's a great question. And on NPR shortwave podcast, we dive into that answer. Listen in the NPR app or wherever you get your podcasts.

B (3:57)

The federal government has shut down. What are lawmakers arguing about and what does it mean for you? The NPR Politics Podcast is here to make sense of it all, giving you updates and news every day to keep you informed. The NPR Politics Podcast Listen every day.

A (4:13)

Hi, it's Terry Gross, host of FRESH AIR. Hey, take a break from the 24 hour news cycle with us and listen to long form interviews with your favorite authors, actors, filmmakers, comedians and musicians, the people making the art that nourishes us and speaks to our times. So listen to the FRESH AIR podcast from npr and whyy it's Consider this from npr. So is the AI boom actually more a bubble? Because Wall street is increasingly afraid of that. In fact, this week, JP Morgan CEO Jamie Dimon said, quote, a lot of assets appear to be entering bubble territory. And bank of America's monthly survey finds that more than half of global fund managers now do believe that AI stocks are in a bubble. So what would happen if that bubble bursts? Well, to talk more about that, I'm joined now by Jared Bernstein. He's a policy fellow at the Stanford Institute for Economic Policy Research. He was also the former chair of the Council of Economic Advisors under President Biden. Welcome.

B (5:23)

Thank you for having me.

A (5:24)

Thanks for being with us. Okay, so bubble or not, it still feels like gazillions of dollars are going into AI companies right now. And we keep hearing how incredibly profitable the chip maker Nvidia is. So just to confirm, there is still tremendous amounts of investment going into AI at the moment.

B (5:41)

Yeah, yeah. And there's tremendous amount of investments going into all the past bubbles we've had starting in the 1600s with the tulip bubble. Right. So one of the characteristics of a bubble is that the level of the investment becomes detached, lastingly or persistently detached from the amount of return or profit that that asset, be it housing or Internet, could Plausibly generate. So the idea that you have a lot of investment flowing in is consistent with a potential bubble.

A (6:12)

Got you. Let's talk more about that because you have recently written an op ed in the New York Times, which you wrote with Ryan Cummings, a fellow economist, and your op ed, it's literally tit. AI sure looks like a bubble. Watch out when it pops. Okay, so first, generally, what defines a bubble?

B (6:29)

It's what I was just saying. Every time you buy a stock, you're speculating on its future earnings. Of course, sure. What happens here is that large swaths of investors just continuously pour more investment into this asset without a ton of regard for how much it could reasonably pay back and by when.

A (6:51)

So let me just make sure I understand. When we call something a bubble, the implication is more about that thing being overhyped in its ability to make money, not necessarily being overhyped as a technology. Correct.

B (7:06)

Critically important distinction you've just made, because I wouldn't want anyone listening to this or reading our piece to think that we are disparaging AI's potential, innovative or economically transformational impact, which could be huge. One bubble we haven't talked about is the railroads back in the 1800s. And same thing, huge investment bubble. It burst, it created tremendous economic havoc, and then it productively transformed the economies that were building it out. What we're talking about is very specifically whether the financing, the level of financing is justified given the amount of returns that it implies. And if it's not, if investors start to get rid of worried about this particular bet, they could unwind that bet. And if enough of them do that, at the same time, then you have a bursting bubble.

A (7:58)

Okay, well, if this AI bubble is indeed a bubble and it bursts, how could that bursting potentially affect all of us? I mean, could it trigger a recession, you think?

B (8:11)

There's no question it could trigger a recession. And in fact, past bubbles have clearly done so. When the Internet bubble burst, the unemployment rate went up a couple of points. That was not as bad as the housing bubble, which led to a shutdown of global credit markets and an unemployment rate in this country that went up over five points. What we worry about in the case of the AI bubble is something called the wealth effect. And that means that if the stock market tumbles enough so that people feel and in fact are a lot less wealthy, they're going to spend less. And real consumer spending has been driving this economic recovery. Should that retrench because of a bursting in the AI bubble and this wealth effect, it's potentially recessionary.

A (9:00)

Well, then, is there some kind of course correction to be done here in order to avoid a burst of a bubble or to mitigate whatever damage results from the bubble bursting?

B (9:12)

The mitigation typically takes the form of fiscal policy, unemployment, employment, insurance, the usual kinds of programs that we implement when the economy takes a hit. There's not that much you can do to deflate a bubble, or at least not much that you can do safely, except what we're doing right now, which is to try to talk about it, to raise consciousness among investors so that the numbers and risks are more transparent versus more opaque. I've always thought that opacity is really your enemy when you're talking about this kind of finance. So I think the more we can be transparent about the valuations that are in place, about the expected returns, about the potential economic impacts of AI, the better chance we have of rationalizing some of this potential irrationality.

A (10:04)

Jared Bernstein is a policy fellow at the Stanford Institute for Economic Policy Research. Thank you so much for joining us today.

B (10:12)

My pleasure.

A (10:14)

This episode was produced by Briana Scott. It was edited by Patrick Jaran Watanan. Our executive producer is Sami Yenigun. It's Consider this from npr. I'm Ailsa Chang. A lot of short daily news podcasts focus on just one story, but right now you probably need more on up first from NPR, we bring you three of the world's top headlines every day in under 15 minutes because no one story can capture all that's happening in this big crazy world of ours on any given morning. Listen now to the Upverse podcast from NPR. On NPR's wild card podcast, Jeff Hiller says fans of his character in Somebody somewhere have a slogan, WWJD. What would Joel do?

B (11:04)

You know, in the 80s they had.

A (11:05)

What would Jesus do? You've been deified.

B (11:10)

Oh, no. This is what brought the Beatles down.

A (11:13)

Watch or listen to that wildcard conversation on the NPR app, YouTube or wherever you get your podcasts. Want to hear this podcast without sponsor breaks? Amazon prime members can listen to Consider this Sponsor free through Amazon Music. Or you can also support NPR's vital journalism and get consider this plus@plus.NPR.org that's plus.NPR.org.