How worried should Americans be as AI threatens jobs?

[0:00] A number of polls, including our own, show Americans are increasingly worried about the [0:04] economy and more pessimistic about where it's headed than at any point in recent memory. [0:10] March brought the biggest jump in inflation in nearly two years. But it's not just prices. [0:15] Americans are anxious about their jobs and specifically about whether AI is coming for [0:19] them. Some prominent voices are calling it catastrophic. Others say it's all hype. [0:24] The data so far is somewhere in between and deeply contested. Let's try to break down some of this [0:31] with Josh Tarengel of The Atlantic. He's spoken with economists, CEOs, and a number of experts [0:37] all about this for his recent piece, America Isn't Ready for What AI Will Do to Jobs. [0:43] Josh, thanks for being with us. My pleasure. [0:45] So there's been a lot of talk in recent months about AI coming for especially entry-level [0:50] white-collar jobs. You've done all the reporting. Paint a picture for us. [0:54] Sure. What I really set out to try and figure out is exactly to your point, like what's actually [1:00] happening and what's about to happen. And so I did a little bit of a tour. I spent a ton of time [1:05] with economists. And economists are divided really in two ways. We have seen technological disruptions [1:12] before. And so economists who love looking backward and comparing data say, you know, there's a large [1:19] school of them who say, look, if this plays out over a decade or more, there's a natural rate of [1:24] adjustment in the labor force. And it may be fine. And it may even be better than fine, because what [1:30] we see is that productivity could lift all boats. And so AI may deliver tremendous productivity. [1:35] There are a cohort of much younger economists, and I point that out because the generational divide is [1:41] really important here, who don't think that their elders are wrong about the data. They think they're wrong [1:47] about the tech. So when you look back at things like electrification, which happened in the early 1900s, [1:53] it took about 40 years to fully electrify America and to see the productivity in the data. The [2:00] difference is that AI rolls itself out. And you're dealing with software that is inherently smart, [2:05] that makes machines very, very smart. And so this younger cohort of economists says, you're not seeing in [2:11] the data yet. But when you see it, it will be too late, because the tech moves that quickly. And we [2:18] will see it in the job force. And it would be too late for us to do anything about it. And so they're [2:22] advocating for making plans right now. What happens if unemployment gets to 10%, 15%. What does a society [2:31] look like when labor is that challenged, and when you consolidate wealth all within one covert of [2:37] people? And so what I discovered is that the economists are really at war with each other [2:42] about what's about to happen. It's a very calm war. It's a very polite war. So in my sort of next phase [2:50] of journey, I went to CEOs and largely the CEOs of Fortune 100 companies. They too are a little bit [2:55] divided. Some have made tremendous investments in AI. Others are a little late to the party. Where [3:03] they're not divided is that they all said to me, look, Wall Street has watched us make these [3:09] investments into AI for the last three or four years. This is the year they are going to expect [3:14] action. And by action, they mean money. And if we don't have gains to show, that does mean we're [3:20] going to make cuts. And we're going to make cuts that we may say are AI related or not. But we are going [3:26] to replace labor with automation. And that was very definitive across a bunch of the largest CEOs that [3:35] I spoke with. I would point out that most of them did not want to speak on the record. And that that [3:40] itself is telling about the state of people's anxiety and ambivalence about AI in the economy. [3:46] Yeah. Well, those younger economists who you say make the point that we need to make plans now, [3:52] we know that's not really happening in Washington. But the the the CEOs you spoke with when they talk [3:58] about smart regulation, what do they have in mind? Well, I think in fairness to those CEOs, [4:04] they are not regulators. Right. And left to their own devices, their job is to maximize shareholder [4:10] value, a phrase that they will repeat constantly. And what they want ultimately is a fair system of [4:16] regulation. Now, they also want the right to say that they don't want any regulation in public [4:20] while clamoring for it in private. But they, for the first time in a long time of reporting, [4:26] I sensed an eagerness among those CEOs for Washington to get involved both in regulating [4:33] the technology itself, but also to be driving contingency planning. Because, you know, a lot of [4:39] these folks, they like their workforces. But the moment a competitor slashes their workforce, [4:44] and their stock price goes up, that competitive CEO is the one whose job is next. [4:51] And so this is what regulation is for. It's to step in when the market can't control itself. And [4:57] what the CEOs were telling me is essentially, they see a scenario where the market will not [5:02] be able to control itself. Efficiency rules all when it comes to shareholder value. So they look to [5:08] Washington. And when I went to Washington, what I discovered is that this is just not really [5:13] something that is on the minds of most mainstream lawmakers. [5:17] Is there a way that people can, in some way, AI-proof their job or incorporate artificial [5:24] intelligence into their work so that it's useful? [5:26] Yeah, look, I think that the AI industry has done a pretty lousy job of educating people about what [5:34] it's good for. They've led with how much money they need to make it work. They've led with job [5:39] replacement. They've led with the downsides of AI. And the truth is that it's pretty easy to use software. [5:45] And it will benefit anybody who starts to play with it and figure out, how can I incorporate this into [5:52] my job? It's not terribly scary when you use it on your own. It's also not flawless. As I'm sure you [6:00] know, like, you know, we've seen statistics that Google Gemini has a nine out of 10 success rate at [6:06] summarizing articles. Nine out of 10 is not actually that good. And so there are still flaws with a lot [6:12] of LLMs. But it's here. It's here to stay. And the best thing people can do to sort of AI-proof [6:18] themselves is to actually figure out, how do I integrate this to make my work faster, better? [6:24] What are the lines around it that I want to draw for myself so that there are things I trust it to do [6:29] and things I don't trust it to do? Everybody's mileage is going to vary based on their personalities, [6:34] based on their jobs. Sitting it out, waiting for regulation, you know, you can't duck and cover. [6:41] This is actually a really important technological moment. And I would encourage everybody to start [6:46] thinking about themselves in relation to AI. Josh Terengal of The Atlantic, [6:50] thanks again for your time. We appreciate it. Thanks, Josh. [6:54] Support journalism you trust. Support PBS News. Donate now or even better, [7:10] start a monthly contribution today.