I want you to fill in the blank for me. Okay. So, I feel blank about AI. >> I feel um mixed feelings about AI. >> I feel anxious about AI. >> It's a lovehate relationship for sure. >> Completely conflicted. AI is amazing and it's making me a better writer, but it's also taking my job away. >> From the New York Times, I'm Zolan Kanno-Youngs filling in as host. This is the daily. As AI becomes more advanced, people are getting increasingly nervous about how it could change the economy and their jobs. >> I'm seeing a lot of job loss because of it. >> It's a threat to my profession. I really have to rethink what I do for a living and probably do something else. >> Oh my god, this thing is going to take my job and not only is going to but actually did. >> But for all the anxiety, what AI is actually doing to the economy remains pretty murky. >> Seems like it's taking over being me. Today, chief economics correspondent Ben Casselman on why AI's impact has been so hard to pin down and what we can learn from the tech disruptions of the past. It's Monday, July 27th. >> How we doing? >> Doing well. >> Appreciate you doing this. >> Yeah. Excited to sit down for this. excited to be hosted by you. >> I'm trying to accumulate as many daily hosts as I can. Like Pokemon, right? >> All right, I'm just going to jump in. >> Let's do it. >> Um, Ben, I am picking up on a lot of anxiety when it comes to how artificial intelligence will impact our economy. We know from polling that about 70% of Americans think AI will lead to fewer jobs. So, as someone who talks to economists every day, how much of your time is being taken up by this question of how AI will impact the economy? I think it is arguably the important question. You know, we talk all the time about tariffs and oil prices and, you know, all of these shocks that are hitting the economy and those are all, of course, incredibly important issues. But I I think it's very possible that if you and I are sitting here in five years or 10 years looking back on this period that the thing we'll be talking about is AI and kind of the early signs of how it was affecting the economy. I don't think we know what that conversation will look like. I don't think we know what big change we will have seen, but it certainly feels like this is the moment where it's all starting. Right. This is interesting because, you know, sometimes I feel like we hear AI is going to be a godsend to workers. It's going to make us all 100% more productive or it'll wipe out all white collar jobs entirely. What do you make of those predictions? So, I think there are a couple of answers to that question. One is that it's early. We're still figuring out how to use this technology. It's still being rolled out. But beyond that, we don't even really know with confidence what's happening now. You know, predicting the future is hard, but even knowing the immediate moment is difficult because our economic data really wasn't designed and isn't up to capturing a change that's happening this rapidly in anything close to real time. Can you explain that a little bit more to me? Why why don't we do? >> Well, I mean, so let me take a simple example of this, right? We get the monthly jobs report and we talk about, you know, how many jobs were added or lost in the economy in a given month. If you go and look in that report the next time it comes out and you want to look at what happened with tech jobs, you will not find that anywhere in the report cuz we don't break out tech as its own industry. These industry categories were established literally decades ago. Tech is kind of sprinkled between a few different categories. Some of it's in the information sector, which also includes newspapers, it includes us. Some of it is in professional services, some of it's in manufacturing. So, you couldn't go and say, "What impact is this having on tech?" You also, >> we're not even talking about AI. We're talking about how government data doesn't even track the tech industry. >> I mean, if you get into the weeds enough and you look closely enough, you can start to tease it out, but there's not like a line in that report that says it.

We've heard a ton about concern about what this means for recent college graduates. Well, we don't reliably track what happens with recent college graduates on a month-to-month basis. And our data infrastructure kind of isn't up to doing that at the kind of speed that we want to know right now. That does seem like a big oversight. So if government data isn't the resource, isn't the way that we're going to get this clarity on AI, where did your search bring you next? >> Well, I mean, so look, first of all, I I want to be careful saying oversight. Yeah, >> the economy changes quickly and we can't just sort of spin up a new measure every time something changes in the economy, right? So there's a reason we kind of track this gradually over time, >> okay? But we are seeing a lot of efforts to use data from the private sector to help measure this in something closer to real time. You know, we have ADP which handles, you know, payroll for many big companies. They're putting out data and economists are analyzing that. LinkedIn, Indeed, other companies that collect and produce a lot of data about what's going on in the labor market. They're putting out reports and economists are diving into all that. The problem is they're all telling different stories. There have been very credible reports out recently from serious economists who have done careful work that show we're seeing losses of jobs for entry-levelvel workers in AI exposed occupations. And you look at that and you say, "Here it is. Here's the canary in the coal mine. Here's the sign of AI starting to wipe out jobs." And then somebody else comes along and puts out a report using good data, careful methods that says, "Aha, the companies that are adopting AI the most quickly are adding jobs more quickly than other companies." So you may have different private sector companies trying to track AI, but they're coming up with takeaways that seem to be a polar opposite at times. >> That's right. Which I think maybe isn't that surprising considering how quickly this is moving and and how quickly this field is developing. But I do feel like there are some companies who are already raising their hands and saying AI is changing my business. >> Amazon has just announced it's cutting 16,000 jobs worldwide. >> It's the latest round of mass layoffs as the Seattle >> Amazon cut thousands of jobs in the past year. They cited AI. Block the payments company is another after the company announced plans to lay off almost half its workforce. CEO Jack Dorsey pointed to AI saying intelligence tools have quote changed what it means to build and run a company. >> Help me make sense of this because I I do think we are seeing these big almost announcements that seem like warning signs from these companies. That's exactly right. Right. We're hearing these big announcements of layoffs in some cases directly tied to AI. When I talk to economists I hear a lot of skepticism about those claims. You know, companies right now are being rewarded by their investors for making big claims about AI, right? Any company that says, you know, we're making big AI investments, we're making big gains, their stock price goes up, right? Their the VC money floods in. >> And so, if you're a CEO right now and and you're thinking, maybe I overhired a little bit a couple years ago, maybe I need to make some cuts. Boy, you are incentivized right now to say not like, oh, I screwed up. I hired too many people. It's oh, AI has made me more productive. >> Oh, this is interesting. So, you're saying it's likely that AI could be a convenient scapegoat for these companies. >> Yeah. Or or look, to be a little less cynical about it, in some cases, AI may be one factor among many. Your business has slowed down a little bit. You're trying to, you know, rethink how you do things and then AI is also creating some opportunities, but it's better to talk about the AI part of that than the other part of that. That's different from sort of saying, "Hey, I'm just going to, you know, lay off a whole bunch of workers tomorrow because AI can do it today." Okay. So, when it comes to measuring AI, government data is outdated. Private sector data is muddy. We can't fully trust the companies. So, what do we know? So, I think we know two things. First, we know this is moving incredibly quickly. We know the technology is developing really rapidly. We know that companies are adopting it really rapidly.

And when I talk to not just sort of the boosters in Silicon Valley, but talk to corporate leaders across industries, when I talk to economists, right, there's sort of increasing confidence that this is going to have a real impact on our economy, our labor force, our lives. >> Okay? And you know, we just actually a few days ago got a statement that was signed by around 200 economists and other experts warning that we need to prepare for this. Because when you actually look at the history of this, it's not that surprising because if we look at the way technology gets rolled out in an economy, it often follows a pattern that economists talk about as a J curve. >> Literally, it's just referring to the shape of the letter J, right? Drops down and then it shoots up following you. >> The idea here is when a new technology rolls out, companies initially have to like try to figure out how to use it, right? We're all trying to figure out how to use video conferencing technology in the early days of the pandemic. We're all trying to figure out how to use the internet when we go back, you know, to the 1990s. And nobody's really sure how to use it. Everybody's fooling around with it. They think it's going to be a big deal, but they don't know how yet. And then at some point, companies start to figure it out. Workers start to figure it out. And all of a sudden, productivity starts to go up. But we may right now still be in this scoop part of the Jay where actually if anything it may be making us less productive in the moment even though it will have this big positive effect on our productivity and and then maybe ultimately on our jobs down the road. This kind of tracks as I'm even hearing you talk because look I cover the White House and I'm just starting to explore with some of these tools and I'm not very good at it. I feel like at times I'll try to like look up an old speech or use AI to form a contact list and eventually I get sick of it and I actually just go back to making calls. But it sounds like what you're saying is we're just at the beginning of this process and maybe over time people like me, even noviceses with technology will learn how to actually use these tools and then hopefully in my case productivity will follow. you start to figure it out as an individual and I'm in the same place on this by the way >> and companies start to figure out >> for [laughter] relating to me a little bit thank you >> but companies also figure out ways to reorganize work they start to say okay you know we don't need as much of this we need more of this new companies pop up that are built around this technology from ground zero and all of a sudden we start to have these real economic impacts that hit once we've sort of absorbed this technology >> [music] >> And that's where we could see big productivity gains, but also where these concerns about job losses and other disruptions start really taking hold. >> So I guess the big question is like what does the rest of this j look like? [music] So I think the best way to answer that question is rather than trying to predict the future is to look back at the past. And I've been spending a lot of my time looking back at one particular period, which is the 1990s. We'll be right back.

Okay, Ben, as a '9s kid myself, I have a high top fade. I love the '9s. I don't want it to end, but tell me why we should look back at this time to understand something like AI. >> So, I'm a few years older than you, I think. I remember >> I'm not going to age you [laughter] then. Yes. So, I I remember the '9s well, and this was a decade where we had two really big consequential forces hitting the economy at the same time. >> It spans the globe like a super highway. It is called internet. It feels a bit like everyday human fellowship, but it's bigger and more precise. >> The first of those is the internet. This was the era of dialup. [music] >> Welcome. >> of AOL. >> You've got mail. >> Of kind of the early days of getting [music] online. It's tapped a yearning to connect to talk with the world about art, music, sex, guitar construction, conservative politics, [music] grief. And it's interesting to talk about that now in the context of [music] AI because we think back on the internet, right? And we think of it as creating this whole new industry. >> The marriage of mobile phone and internet technology which is working the world's stock markets into a fever of anticipation. >> Setting the stage for tech as we know it. Setting the stage eventually for the mobile web and iPhones and apps and all of this, right? every business, no matter how large and no matter how small, will be on the internet in the year 2000. >> It created all these jobs, but it also wiped out lots of jobs. >> Goodbye. We used to have typists. We used to have typing pools and and then later word processors right before word processor was a program. It was a job. We had whole categories of jobs that have been wiped out or dramatically reduced. We don't have nearly as many travel agents as we did before the internet, before we could all go and book our flights on Orbits or Expedia or whatever it is, right? We had to go to the bank a lot more before we could do it on our phones. We eliminated a lot of jobs, but we don't remember it as this mass job losses. >> Why is that? >> Because it happened gradually and it was diffused across the economy. This is not a story where one day they walk in and they fire the whole typing pool. It's not, you know, one day we we stopped having travel agents. It was that as companies figured out how to use this technology in different ways. They started to change these jobs. People were later in their careers had time to sort of wrap up their careers and retire. People who were earlier in their careers had time to learn new skills or to change direction, right? To say, "Hey, maybe this isn't the career that I want to go into. maybe I should go in a different direction. >> You had time to pivot. >> You had time to pivot. You had time to see the writing on the wall and to say, "Hey, maybe I should go to college because that's the direction that the economy is moving in." It wasn't like every company in this town all shuts down at once and now where am I going to go? >> It was spread out and gradual enough that people had an opportunity to react. And so we don't have these mass job losses. We have job shifts that people are able to react to over time and to pivot into new areas that have more opportunity. And we look back now and we remember this as this period of growth and opportunity even though it was a period of tremendous disruption and uncertainty in the moment. Okay, this is fascinating. It seems like what you're saying is the reason we look back on the internet revolution as not a time where jobs were lost and entire fields were made extinct. The real reason for that, the real factor seems to be time. >> It gave them time to adapt. >> It gave them time to adapt. So now the question would seem to be just how fast is the AI transition going to play out and will it be so fast that people can't pivot. >> I think that's exactly a question and it is what we'll learn over the next few years but we don't know the answer to that yet. >> Ben, it does seem to me particularly as we sort of compare these two periods like the internet was a tool to workers. It's a tool to us now, right? But when I talk to people in Washington, when I talk to just friends around the country, they're not worried about this as a tool. They're worried that they could be replaced by AI. So, how do we make sense of that as we compare these periods? That is the fear. I don't think we know yet that that's the reality. >> Okay. Okay.

AI at this point, for the most part, is a tool. It's a tool that workers are using in different ways. they're finding it more or less useful, but for the most part, right, you're talking about starting to explore using AI in your work, right? But there's not there's not an AI White House reporter right now. And so, we don't yet know whether AI is really going to take my entire job or somebody's entire job. >> I don't think AI could ever take your job or my job. Just going to put that out there. But don't let me interrupt you. As we go through this part of the Jay, as companies and workers start to figure this out, we'll get a better sense of whether does it actually replace some of these jobs or does it just make all of us more productive? And I think that is sort of the core question right now. >> Okay. I will say thus far you are doing a great job of explaining this. Not a great job in lowering my blood pressure and anxiety levels on this topic. But >> that was the good news conversation. We haven't even gotten to the bad news example. >> What's the bad news example? So, this is the second big force in the '9s or or starts in the '9s and picks up steam in the 2000s, which is the so-called China shock. By adding China to the WTO, we strengthen the organization by further integrating China's 1.2 billion people and 1 trillion dollar economy into the world market network. This is this period where we open up trade with China of course and all of a sudden we start to see this rush of competition that leads to huge job losses in particularly kind of the southeast and the Midwest. [music] >> And here's what we know right now. Thomasville Furniture's plant C will close by mid July. The company will move that plant's production to a factory. >> Crews demolished what was left of the old Pillowex plant early this morning. It took more than a ton of dynamite. Mills announced this morning it is shutting down two of the county's four plants. The closings at plants in Cliffslide and Florence mean the loss of hundreds of jobs. >> But I think the critical thing to recognize there is that this was a fast shock. [music] >> We go back to time. >> We go back to time. >> More than 800 jobs have been lost and that's just since December. This is where we see >> entire factories [music] close up, entire industries shut down in a matter of months or years. >> My whole family worked here and I got a job here and everything is it's a shame at shutting down. >> Right now things don't look too good. Unemployment rate in the county will probably approach eight or nine, maybe even 10%. And as I think about this time period, Ben, I'm also thinking about entire sectors that just collapsed, right? You mentioned before how the internet revolution not only was gradual, but it was broad. I'm thinking about the town in small town America that relied on one factory and during this time period, that factory went down and hundreds if not thousands of people lost their jobs and entire communities basically collapsed. I mean the ripple effects were really severe during this point in time. >> That is exactly right. It was fast and it was concentrated. >> So take Hickory, North Carolina. This was a center of furniture manufacturing right in the 20th century. Well, we open up trade with China and we get flooded with cheap furniture from China. Hickory can't compete. The factories shut down. We see tens of thousands of jobs lost just in the hickory area in this one industry. >> Tens of thousands. >> Tens of thousands. And what happens when that hits in one area? Well, think about all the other jobs that depend on that, right? Those workers are shopping in the retail stores in the area, right? They're sending their kids to the schools in that area. They're eating at the restaurants in that area. All of those industries get hit as well when we get this kind of concentrated shock. You can't even move out of the area because who are you going to sell your house to? >> That's right. Why would I want to buy that home if I'm going to an area where the main industry that the community relies on has collapsed? That's exactly right. And so we see this in Hickory, but we see this in communities all across the country that has these really powerful impacts and these lasting scars. And Ben, I think we've reported on sort of either sides of this, right? The stakes of this moment are really severe. We're not just talking about economic impact. We are also talking about communities that had addiction levels rise rooted in that unemployment problem. This brings us to the opioid epidemic. We're talking about also an impact on our individual workers but for the economy as a whole with all of the kind of social and political implications that we were just talking about.

You know even with the uncertainty we established at the top of this you have by bringing us to the '90s outlined that there are some some lessons from significant disruptions in the economy that we can draw from. We can draw lessons from the past. So, are policymakers acting on those lessons? >> I think policymakers are starting to grapple with those lessons, which is not the same as saying that they're acting on them. You know, we're starting to kind of see this process play out both at the political level and at kind of the Washington think tank policy level. We're starting to see some discussions in Congress and in state capitals, but I don't think we've seen anybody from either party kind of lay out a comprehensive plan that anybody thinks is really going to tackle this in a big way. >> I think many would find that concerning. Uh I mean what should we be doing? When you talk to economists, what do they say? >> Well, so when I talk to economists now, what I hear from them are a few sort of things that we really need to be tackling. One of them is this measurement question. Can we improve our ability to track this, right? Can we develop new tools to allow us to measure this in more real time so that we actually know what is happening so we know which workers are being affected and where they are and what is happening to them. >> Identify who may need help the most and then go and help those people. But first you got to identify which sectors are going to get hit. >> That's exactly right. Then the question becomes okay now what do we do to help whoever is being affected here >> right >> when I talk to economists and policy experts on this one thing that they say is look given the uncertainty step one is to shore up some of the existing systems that we have you know the US has an unemployment insurance system but we all learned during the pandemic how rickety that unemployment insurance system is how >> many struggle to navigate that system and and needed it and relied on it. >> It was incredibly helpful for a lot of people, right? But it was also a fundamentally pretty broken system in a lot of ways. We developed in the 1990s and further back trade adjustment assistance that was meant to help workers who were displaced by globalization, but it never really reached a lot of the workers who needed to benefit. So there are a lot of policy experts now who are actually looking back at the lessons of that period and saying like how do we make sure that we design programs better this time around and then there's you know do we need to have some sort of totally new program that deals with AI specifically right maybe that's the government taking stakes in companies to create you know some sort of sovereign wealth fund that allows everybody to benefit from AI maybe this is some form of universal basic income where instead of working people are getting checks directly from the government in this new world. None of that is anywhere close to an actionable policy right now. But it shows you sort of the extent to which people in Washington, people you know in capitals frankly around the world are starting to grapple with sort of the potential impacts of this technology and the possibility that maybe we need to take a totally different approach to policy.

Okay, so we've been talking about what policymakers should do, but those listening might be wondering what can they do, right? What is the individual supposed to do? For those with kids, what are they supposed to tell them? [music] I think that that in many ways captures sort of what is scariest about this moment. If you think back to that 1990s period, there was a sense that you did know what you should do. You should go to college. you should pursue these new burgeoning careers. And look, that did not work out for everybody. We know that. >> Yes. >> But there was some sense that this is the direction the economy is moving in. Go that way and you can do okay. We don't have an answer now to that in the same way. There are no doubt going to be new jobs that are created through this AI innovation, but we don't know what they look like yet. I don't know what to tell somebody to go and major in today or even whether to go to college or not to go to college, right? Those calculations are changing in ways that we don't fully understand. And because we can't give clarity, it's inevitable that people are going to really feel like they don't know what they should do and where this is headed. Well, it sounds like for at least the time being, we will have to embrace the uncertainty. >> I don't know that we have much choice. >> Thank you, Ben. Thanks so much for having me. >> I think AI has the chance to revolutionize everything. We're not ready at all. >> You know that one guy from Office Space who says, "I'm a people person." We're all becoming that. I think we're just taking specifications from business people and feeding it to the AI these days. 6 months ago, I'm not wasn't sure when I was going to retire, but AI has actually accelerated that a little bit because things are changing so much at work. There's a part of me that's like, you know what? I'm ready to to end right now. I've got grandsons who are in in fifth and seventh grade coming up and I wonder what they're going to be doing when they get to college. Are they going to, you know, are certain choices going to be gone? >> Well, in Brooklyn, we say, "What are you going to do? You You got to It's It's here. We'll be right back.