Moody's Talks - Inside Economics - AI: Friend or Foe?

Episode Date: February 27, 2026

Mark, Cris & Marisa reunite for a lively discussion about their predictions around AI’s impact on the economy over the next year or two. The team talks about their recently released webinar & white paper on the Macroeconomic Consequences of AI and answers several great listener questions in the process. Marisa and Cris try to talk Mark down off the AI-apocalypse ledge, as the once eternally optimistic Zandi has gone down a darker path recently.  Jenna Score: 8.5/10 For a deeper dive on AI and the macroeconomy, see our new paper, The Macroeconomic Consequences of Artificial Intelligence, where we model four potential economic paths over the next decade. We also walk through the scenarios in a companion webinar available now on-demand. Read the paper: https://www.economy.com/getfile?q=2B555C90-1118-4A49-BDAA-5C0A99F83A9E&app=download Watch the webinar: https://bit.ly/3OF6dn9 Read the Citrini Research Scenario on AI here: https://www.citriniresearch.com/p/2028gic Email us at [e-mail] for more info about the Moody's Summit '26 Conference in San Diego Hosts: Mark Zandi – Chief Economist, Moody’s Analytics, Cris deRitis – Deputy Chief Economist, Moody’s Analytics, and Marisa DiNatale – Senior Director - Head of Global Forecasting, Moody’s Analytics Follow Mark Zandi on 'X' and BlueSky @MarkZandi, Cris deRitis on LinkedIn, and Marisa DiNatale on LinkedIn Questions or Comments, please email us at [e-mail]. We would love to hear from you.  To stay informed and follow the insights of Moody's Analytics economists, visit Economic View. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Welcome to Inside Economics. I'm Mark Sandy, the chief economist of Moody's Analytics, and I've got a full house. My two co-hosts are with me. It doesn't happen very often these days. Both Christorides and Marissa DeNatal. Hi, guys. How are you?

Doing well, Mark. It's all in the family. Yeah. Is it just me, or it's been a while since we've been together? No? All right. I was not on last week, but. Yeah.

The week before, I think you were. Were you? Oh, okay. Well, it's good to have you all here. Yeah, I'm actually in Orlando today. I came over from Vero for a University of Florida Real Estate Conference, Commercial Real Estate Conference and spoke at that.

It was a nice experience. And then I'm headed over to Tampa because I have a board meeting over there. So making the trek from the east coast to the west coast of Florida. And it's, I'll have to tell you. kind of nice guys kind of nice okay yeah no stops at disney no uh well i almost did that by accident right you know i was coming to the to the resort to the hotel and uh i was on my phone talking and not paying attention and all of a sudden i look up i'm in epcot i go what's that

i didn't really mean to be there yeah but i made my way over i made it here so all good um but i'm Head it back up north in a couple weeks. Yeah. Okay. Should I come? What do you think? Yeah, it's warming up now. It's warming up.

Good, good. Well, it's good to have you. We have no guests. So this is just the three of us, which, that feels good. We haven't done that in a while. And I thought we'd spend most of our energy on listener questions. Marissa said we had 45 questions in the queue. So we're going to go through those.

before that though I thought we talked a little bit about AI we have a paper out and we thought we might talk about that because AI has been in the news obviously and then even before that I thought we talked a little bit about the Moody's Summit that's coming up for you guys you guys are going to be at the Moody Summit right we are we are it's in your backyard it is in my backyard yeah San Diego yeah I think it's May 5th and 6 That's right. Yeah. And this is the kind of the big event of the year that Moody's puts on. A lot of financial services, banking. And we have an inside economics day at the summit on the sixth. And we'll go over a lot of pretty cool economic issues and topics. What are you talking about, Chris, at the summit?

I'm in an affordability session. So we're talking about K-shaped economy and all the different affordability issues. And are they real? Are they over-emphasized or not? And then also in a consumer credit session with you, actually, Mark. We have a panel on consumer credit trends with our colleague, Mike Brisson. Oh, am I on that panel too? You are.

Oh, cool. Cool. You know, have you noticed, just as a quick tangent, that the number of subprime borrowers is up quite a bit in the last, looking at originations. If you look at originations, you know, like at bank card, for example, the number of subprime borrowers that are taking on new credit card debt has risen quite significant. You know, it's not off the charts, but it's noticeable. Did you notice that, Chris? I did. Yeah, things are, it kind of points to the K-shaped economy as well, right? Just that

you have certainly a lot of consumers that are struggling, that are having to, you know, to take on additional debt. And they're also seeing their credit scores being impacted by some of the higher delinquency rates, student loan debt now, skyrocketing in terms of its delinquency. So there's definitely some stress. Some stress in a two-track economy here that's going on.

Right. Yeah, of course, I'm referring to all the credit file data we get from Credit Bureau Equifax, right? And, Marissa, what are you speaking about at the conference at the summer? AI, artificial intelligence. Excellent. Yeah, me and Dante.

You and Dante. Dr. Di Antonio. That's right. Cool. Are you the bull and he's the bear? We haven't decided if we're going to do that quite yet. Like you and Dante like to do. Good cop, bad cop.

Yeah. Yeah, we're going to talk about, yeah, we're going to base it on the paper that we just wrote this past week. Yep. Right. Right. And if you're, you know, obviously interested in all the, you can take a look at the agenda, but if folks are interested, inside, contact us and we'll hope to see you there, Inside Economics at Moody's.com. So please, feel free to come and join us in San Diego in a couple months. And talking about AI and the paper, you know, it was a week of papers. Did you see that, I think it's Centrini Research kind of a scenario came out? Yeah. Did you see that, Chris? Yeah, pretty dire.

Dyer. Right. I was going to say scary. Yeah. That's the kind of that it was one of our, in our paper, what we did is we examined the impact, the macroeconomic impact of different scenarios. I think it's pretty clear the uncertainty around how AI and what it means for the economy is going to unfold in potentially in lots of different ways. hard to handicap, and therefore we examine different possibilities, different scenarios. One of the scenarios is a pretty dystopic, scary kind of scenario where you see this upheaval

in terms of the job market that the productivity gains from AI come on so quickly that it kind of overwhelms the economy, doesn't have time to adjust, that you have a lot of job loss, but you don't see kind of offsetting job gains in other sectors. It just happens too fast. That was the scenario that Centrini kind of focused on. And we included in our paper as well, not nearly as dark, I don't think. Ours was more quantitative than the Citrini analysis and hard to know what numbers they'd put on the scenario. But nonetheless, very qualitatively in the same direction.

Yeah. What did you think of the, did you read the Centrini piece of Marissa? Yeah. So it was more of an imagining, right, of the future, kind of a story about what the future could look like. It wasn't, you're right, it wasn't as, it didn't seem to be as quantitative or, or model based. It was more of a dark landscape. Right. I guess, right, where AI just sort of embeds itself into all aspects of life and jobs and creates massive unemployment. and people's skills atrophy, kind of our tail scenario, I guess, right? The one we attached the lowest probability to. Well, no, no. Well, you know, it's towards the tail, but it's not, I wouldn't call it a tail scenario.

I mean, we attached to 20, I believe, correct me if I was it 20? 20. 20% probability. We called it the job market upheaval scenario and attached to 20% probability to it. So that means there's a, a pretty considerable amount of tail out there where you get some scenario that's in

that spirit, you know, where you get significant job loss, increasing unemployment. So it's certainly not the most likely scenario, and it is kind of toward it on tail, but it's not inconsequential probability. It's something one in five probability is not inconsequential. It's something you have to consider. That's true. I mean, I guess all our probabilities are pretty high. We don't have any real, real tails, right?

Like even the, I think our lowest probability, the productivity boon was, was it 15%. Which is still pretty high for. That's on the other end of the distribution. That's where things go outside. Exactly right. You get all the productivity gains, but you generate enough income and wealth that it creates enough consumption, creates enough jobs elsewhere that you kind of end up in a reason a good place on the job market as well. But yeah, it's a pretty flat distribution.

You know, the, the, the, even the baseline scenario, which is a relatively sanguine scenario, uh, was a 40% probability. So that kind of gives you a sense how flat it is. Yeah. And goes to the uncertainty around it, around it all. So, so Chris, in, in your heart of hearts, uh, that job market upheaval scenario, that dystopics scenario, the kind of the, Centrini research scenario, do you think 20% probability is the right probability? I mean, I'll have to tell you my feelings about what the probabilities are very minute by minute, depending on exactly what I'm looking at or hearing. Like, for example, I heard Block, you know, the FinTech company announced yesterday,

Jack Dorsey is the CEO, that Block is going to cut its workforce 10,000 people to 6,000 people. And that was because of AI and the capabilities of AI. And I listened to that and I go, oh, my gosh. Do you have the same kind of feeling I do that that probability is all over the place, depending on what you're looking at? It definitely varies throughout the day. For me, it depends a lot on what I'm doing. So I did this analysis.

I updated some analysis yesterday. I was using AI tools to do it, and it was just amazing, right? Something I had done last year, suddenly is able to rip through it using this AI tool, and I'm thinking, wow, this is it, right? This is taking on my job.

So, yeah, definitely I could see how that anxiety creeps in, and my odds changed throughout the day. But then, you know, there are other parts of the day. or I talk to others where I, you know, there are still many jobs that are that are very AI resistant, I would argue. They still require physical presence. And so, you know, I don't, I'm not buying into the doomsday scenario quite yet. Also, I think that we still have the capacity for new ideas that this is going to be a great enabler. And we just have to think a little bit more creatively about what's on the other side of the mountain here.

Right. Right. Once you have this tool, everyone has this tool at their disposal, I think we'll come up with new services or goods or things that we, that's a human capacity for imagination that will continue. Well, you know, it's to some degree it's a matter of timing, right? Yeah. I mean, at the end of the day, it's hard to argue against productivity. I mean, you want productivity gains, right? because that's the magic elixir for improvements in people's standard of living and higher incomes,

higher wealth. But it does feel like there's a real possibility here that before we get there, we've got all this hard, a lot of difficult times ahead of us in terms of what's going on in the labor market and the job market. So 20% probability on that scenario, it feels okay to you? feels you still go with that on our scenario or on the trini scenario was that scenario on steroids and right that felt like that probably was out in it I'd put that on the tail yeah way out on the tail maybe a I don't know 5% probability 1% probability something like that because that's a really dystopic version of the job market of evil scenario that we've been talking about

that's right that's right yeah so 20 15 20% is what I would assign to our you would say Okay, okay. So on the optimistic side of the, if you can characterize it that way. What about you, Marissa? That 20%, does that feel right to you? Or how, what do you think about that? Yeah. Yeah, I think I'm right in that range as well. I'm more of an, I think I'm more of an AI optimist than, certainly than you are, Mark. Yeah. You've taken a real dark turn these past few weeks on AI. Have I? You sensed that, have you? Yes. Yes. A lot of talk about.

singularity in the matrix and it started with the Maltbook, right? That was Oh yeah, that's right. That's kind of down a bad path, right? Yeah, I'm a little more optimistic about it. And I guess I just look back to past technologies

for some grounding that we haven't really had a scenario that's been dystopic because of a technology adoption. It tends to be a lot slower it tends to be a lot more stop and start and figure it out, to give companies and workers time to sort of recalibrate.

And we've also seen that, like, think about the internet, right, that there are just jobs that get created that we can't even envision what they're going to be. I mean, we can throw out some ideas of what jobs could be created, but I just think there could be a whole infrastructure and industry around things that happen because of AI that we can't even envision. No, I agree with all. That's all right. That's all true.

But it goes back to the question of timing. Probability? No, the timing. Yeah. All that's going to happen. I feel reasonably confident, but I don't think it happens in 2026 or 27. That's a process. It just feels like it feels like the dark side, meaning the job loss that's coming,

is just going to happen. is just happening a lot more quickly than those other technologies. And it doesn't allow the economy time to adjust and for wealth to be created and incomes to be created and demand to be created and new jobs to be created to offset the job losses that are entrain right now. Block being the example, you know, the most salient example at this point in time. So that's what makes me worried. I mean, like I'm using Claude now to.

So here's an example. You know, we have these current quarter GDP models. The model takes incoming monthly data, weekly data, uses that, applies some econometrics, and comes up with an estimate for GDP growth in the current quarter. So it's a way to kind of summarize how the economy is doing in a more real-time way. Q4 2025, the current quarter model of predictions were really all over the map, you know, more than I've ever seen or a very, very unusual dispersion in estimates. I mean, I think like the Atlanta Fed tracker, which is a current quarter model that people use, was close to 3%. And at times during the quarter was as high as 5 or 6%, you know, something like that.

Our estimates were a lot lower. I saw some estimates that were in the ones. And the reality was it came in at 1.4%. So I look at that and I say, you know, why do we miss and can we do better? So I use Claude. And I built my own current quarter GDP model. And, you know, I'm still learning. I'm still figuring it out.

But the thing is so useful in back testing and swapping out variables, asking it which variables it thinks is more important. What are the limitations to the model? Where should I be focused and try to prove accuracy of the model? And it's giving me suggestions and we're trying them out on the fly. And I'm coming up with a model that I think, you know, this is pretty good. And I'm going, how many man hours did we spend on that building that GDP? current quarter model. A lot over the years, a lot. And I'm here doing it on the fly. And what does that

mean about the main hours that we need in the future to do that kind of work? So you had Claude build the model and choose the variables and stuff. Yeah, I gave it, you know, I had a conversation with it and asked what else do you need? Where's limitations? You know, what's working well, what's not working well, am I actually defining what is well mean? What does that mean? How do you measure that? Am I doing that? Do you have any suggestions around how I should improve that? And it was asking me all kinds of questions, too. He goes, you know, do you have any suggestions about the government's, because one of the misses and current quarter model for 2025 Q4 was we missed the decline, the big declining government spending related to the government shutdown? And the reason was we have,

no variables in our model that captures that. We go to the Treasury statement, but that Treasury statement did not pick that up. Therefore, we did not pick it up in the current quarter model, nor did anyone else. So, you know, what kinds of variables should I be looking at? This is Claude talking to me to be able to, you know, try to pick that up in a better way. So, you know, that's a concrete example. And, you know, there's a big difference between what I've done in operationalizing that and getting it up on our economic view website and doing all the things that are necessary to get it across the finish line and get a lot to be updated on a regular regular basis but nonetheless you know that advanced the ball to you know a significant degree

and what does that mean about the you know the the kind of uh human resource i need in the future to do that kind of work chris what does that does that what do you think yeah certainly but i think that proves my point It all originated from you, though. You had to ask the question. You had to. You had the idea, right? And my point is, I don't think we run out of ideas now that you solve a current quarter model problem, let's call it, if indeed that's done.

It's not like you sit on your hands now and say, well, I guess there's nothing else to worry about this is an enabler. Now you're going to say, oh, well, maybe I can do this monthly. Maybe I should have a daily version of what other data should I be gathering, right? you're going to come up with more and more ideas. So I see this again as an enabler. And that's what I can think of. And on the other side of this,

I think it just opens up other possibilities to answer questions that we've kind of put to the side for a long time. Yeah, but again, I agree with all of that. But what happens in the immediate near term to the folks that, you know, the lesser skilled folks that would be executing on that idea, I don't, I don't, we don't, we don't need as many of those resources, right? So what happens?

And that all, that all happens very quickly. That happens now. It's happening now. You know, no? Well, maybe it is happening now, but I still believe there is diffusion that takes time here, right? It's not like legacy companies just decide, oh, we can do this, adopt this new technology and forget everything that we've been doing. these things, there are some physical limitations here. There's still some skills and whatnot that

have to be adopted. Organizationally, we have to change how we do business around this new capability, right? So I still think that the diffusion curve is out. Maybe it's accelerated, but I don't, I don't think it's instantaneous. I still think there's going to be sufficient time to allow this technology to diffuse. And on top of that, I always come back to demographics. graphics, right? We actually need this technology now. We need that additional productivity enhancement. It's going to that those demographic effects. Because aging at the workforce, they're aging out, and the immigration way down, birth rates are way down, we're going to feel the real pinch here, right? Before AI was on the scene, we had been talking about population and, you know, the implications

of long-term population trends for economic growth. And now this is suddenly a tool that perhaps can offset some of those headwinds we're going to be facing. Yeah, I'll say one again, just... The timing. It's a timing. And the current environment that we're operating in, the job market is not creating any jobs. Zero. We're not going anywhere.

Do you attribute that all to AI, though? No, no. But that's the initial condition. We're starting from that point. And that's a function of lots of stuff, you know, economic policy, immigration to the supply side of the labor market, tariffs and trade on the demand side of the labor market, you've got doge cuts, you know, all those kinds of things have contributed to that. So no. In fact, I would argue very little of what we've observed in the labor market over the last year, the flat labor market at best that we've had over the past year is related to AI. If anything, it's actually lifting employment, right? Because the demand side effects of AI are been much more, they're very positive, investment spending and the wealth effects on consumption due to the run-up in AI stock prices.

It's generated a lot of demand for labor. And the productivity effects that they're just starting. You can see in the tech sector with coders, you can see it a little bit in hiring rates, I think. but you have to really squint to see it. So AI so far, at least in 2025, up to this point in time, was more of a job creator than a force that's limiting job creation. But here's the thing.

That's changing very, it feels like that's changing very rapidly. And the demand side of stuff, that probably stays positive, although questions about that in the context of AI stock prices and whether they continue to rise and whether they continue to rise and whether we continue to get those positive wealth effects. But putting that aside, the supply side effects of AI, they're going to kick in. And the question if they start to kick in in a significant way in 2026,

in a world where we're already not creating any jobs, doesn't that result in something that's going to be pretty difficult to digest, and that's job loss? Here's the other thing that's why this block announced me, I find a little unnerving is because once one company, and let's call it block, you know, says we're laying off and is very explicit about why they're laying off, it's AI. I can do this with, you know, almost half the people that I had before. Then other companies start to lay off.

It's, you know, that's the way layoffs happen historically. It's one company in each industry that says, I'm going. And then everyone else is right there behind them. And we start to see, you know, layoffs. And that's the only thing that has kept the labor market from losing jobs. It's the lack of layoffs. They were up to this point of remained very low. But if we get any layoffs because of AI, and again, it feels like that's got to be

dead ahead, you know, given all the things that are going on here, doesn't that mean we get job loss in, in 2026 and 2027? What do you think? Yeah, we could. Talk me back from the lead. Yeah. I just, is it going to be? again, we might have seen job loss anyway, given the hiring that went on during the pandemic,

some of the structural changes that are going on, all the policy uncertainty, right? And you're arguing, well, AI is going to be this great accelerant, right? It's going to really start to lead to the layoffs. And I'm not so sure. I think there's still, if we think about industries broadly, there's still a lot of changes that need to occur. Maybe a technology company like Block where the AI is front and center. It can do a lot of those tasks very efficiently.

But I don't know. In auto companies, I think they're using AI. They're going to continue using AI. But I don't see that they can. Well, think about all the bad office stuff that these guys still have. Like an automaker. I mean, they're still, you know, they employ a lot of accountants. Yeah.

A lot of it's outsourced. A lot of it's overseas. So I might be actually more worried about overseas employment, right? Right. From that perspective. But I don't, I guess I don't have quite the pessimism that you do, that this is ripped the band-aid off. And I don't oversell it because if you didn't ask me what my probability was, I mean, our papers. I'm scared.

I'm scared to ask you. You're scared to ask. It's like, this is happening tomorrow. Yeah, yeah. Yeah. No, I mean, I'm, I still, it's not my baseline. I don't think it's my baseline. I think we still navigate through,

but I just don't say that with any confidence. And increasingly, you know, I think there's a possibility, the odds of us going down that job upheaval scenario path, that Centrini Research-esque kind of scenario just feels like it's not inconsequential. It's over 20% and it's on the rise. It's not 50%. It's not my baseline, but holy cow, I mean, it just makes me very nervous.

Marissa, where do you stand on? I mean, given you just heard what I laid out, you're clearly on the optimistic side of this. How would you push back on what I just said? It almost, there's almost a feels of, it almost feels, bimonal. I was going to say inevitable, increasingly. Oh, yeah. Maybe, maybe my probabilities are over 50%. I don't know.

But anyway, how would you push back on what I said? I think what you're saying, definitely. applies to the technology sector. And that's mostly what we've been seeing so far, right? Is big tech companies block as a fintech company? They're obviously on the cutting edge of incorporating AI into their business. They've probably been doing that to some extent for years, and they've accelerated that. So I could see kind of a bloodletting in the tech sector, which, by the way, we see all the time. Going back, 10 years, we see huge layoff announcements coming all the time in the tech sector.

They tend to scoop people up during economic downturns or weakness. They get really bloated, and then things start to soften, and they let a lot of people go. And they give a lot of different reasons for doing that. So I buy it in the tech sector. I'm less convinced that this is going to happen this year or next year in other kinds of kinds of businesses that aren't as positioned or organized to really incorporate AI into their day-to-day processes. It's kind of like the conversation we were having before we got on the podcast, right? We're all using AI, but how many companies have actually embedded it into their

business processes so far? Tech companies clearly are. I'm less convinced that other kinds of businesses have done that at scale that would lead to the kind of job loss you're talking about in the next year or two. Eventually, they will. Eventually, there'll be consulting companies that do nothing except come into your company and set up AI systems because your business isn't poised to do it, right? That'll be like, I think that's a no-brainer business or industry that's going to come out of this. So, but that's going to take time for companies that, aren't poised to do that in the next year or so. Tech is doing it because they're the ones building in many cases,

these kinds of systems. They're really poised to do this kind of thing. So we could see huge tech layoffs, yeah. But is that going to translate to all different other kinds of businesses and manufacturing, construction, trade, professional services? Maybe at the margins, but I don't see it being very, very, diffuse over the next couple years outside of tech. Anyone look at betting markets?

Is there, have you looked, Chris, I know you, you're a gambler, a bit of a gambler. Have you looked at betting markets? I'm sure there's a question out there that, you know, goes to this, you know, is this what's the probability some Trini research kind of scenario unfolding here? Have you noticed? I haven't seen a, take a look. I'm sure there's a market out there. I'm sure there's some market for some type of question out there.

But there has been some pushback on this Trini research, right? There was, and kind of in line. So what do you think about the Citadel came out with a piece that said, look. Oh, I didn't see that. If you look at Indeed job postings for software engineers, they're actually up 11% year. Yeah, but isn't that AI? I can't believe that that's not AI. But is that displacement from, on the one hand, we're cutting at block.

But on the other hand, maybe the automakers are scooping up some of the, those software engineers because they need to automate their processes, right? So, again, as this diffuses through, you can see displacement in one area with growth in an other area. Right, right. And entrepreneurs, I guess I'm still pretty optimistic in the entrepreneurial spirit. Someone's going to figure out how to use this for all sorts of things. Right. Of course, the other scenario, downside scenario, we considered, was kind of a, it is rooted in history, the Y2K bubble, that the stock market, AI stock prices have gotten ahead of themselves.

And you can kind of feel it in the markets this week, right? The stock market, you know, when Nvidia came out with, and you look at the numbers, you go, oh my gosh, this is a joggernaut in terms of revenues and everything. Yet the stock prices have stumbled, have declined. and that goes probably in part to the fact that investors expected all that good news and expected more and didn't get it and thus the sell-off. So that gets to the other scenario where, you know, investors are just,

AI falls flat. That's the scenario. That's the name of what we call that scenario where AI falls very short. It's still a significant technology. it has tremendous impacts on productivity and profitability of companies, but it falls short of these lofty expectations of investors. And you see this sell-off in the equity market. And that goes back to the wealth effects and demand and what's been driving the economy. That feels equally almost, you know, worrisome. I mean, and it's almost diametrically opposed, right?

One, the Centrini-like job market upheaval scenario is, you know, AI comes on so fast, the productivity gains are so significant that the economy is overwhelmed. The other is the opposite, whereas it doesn't happen as fast, the businesses don't adopt. It just took what, what Amrissa articulated and said, yeah, that's what's going to, that's what happens. And you see the sell-off, and the economy struggles as a result of that. And that's what happened kind of is, Y2K-ish, right? I mean, you saw the run-up in AI stock prices.

Things didn't live up to expectations. You saw the bubble burst, and that led to a recession that followed. So that's an, and I think, Chris, if I recall, you put a higher weight on that downside scenario than the job of peevil scenario. Yes. Yeah. Yeah. Yeah, that's a pure bubble.

Pure bubble. scenario, right? We've just gotten ahead of ourselves in terms of the evaluations here and the earnings potential. I think that's very likely. And it's not to say that AI doesn't pan out eventually as a technology, right? Just like the Internet. But that we're being overly exuberant. And maybe, maybe in the next couple years, you know, investors are expecting this all to pay off, right?

But companies flounder. They don't, they, yeah, we got all these tools, but we don't. Our workforce doesn't know how to use it. They're overwhelmed. They're actually wasting a lot of time doing things that aren't panning out. And you don't get the productivity gains that are implied by all this investment. I think that's a very likely, I mean, after the baseline, that's, I think, the next probable, highly probable scenario. So the job upheaval scenario, the collective of the group said 20%, you guys say,

2015 you're on the optimistic side of that. The group said the AI falls flat scenario where the stock market stumbles and takes out the economy because of the negative wealth effects. We attached collectively a 25% probability to that. It sounds like you guys would attach a higher probability to that. 30, 35, yeah. Yeah, I think I'm initially higher on that too. Right.

So can I just point out that if you take those two scenarios, and maybe I can't do this exactly statistically. It's over 50%. Can I just point that out? Yeah. Yeah, that's right. That's right. So you guys are, you're like saying, Mark, you're too optimistic, but you guys are

pretty, I don't know, isn't pessimistic. Well, I'm not up. I'm more optimistic on AI. I'm less optimistic on the stock market. The stock market. Yeah. Right. Anyway.

All right. Okay. Well, I'm sure we're going to be talking. a lot. Oh, in fact, we have a few guests coming on. It's premature to say who, but we have a few guests coming on that are well-versed in this area coming from the tech industry, the AI industry, to help us kind of navigate through all this, because I think this is going to be, obviously, a big deal for quite some time to come. And that paper we've been referring to, it's called the macroeconomic consequences of artificial intelligence. We'll put that in the show notes,

but you can find that out there on the web. And I don't know if we can do this, but you might want to include the Centrini research and link to that as well because we've been referring to that and people might find that of some value. Okay, let's move on. Let's take some listener questions.

Unless you guys, we should talk about something else. Anything else? Before we go to the questions? No? Okay. Well, let's go to the questions. What do you got there, Marissa? So we have a ton of questions about AI, but I feel like we kind of answered a lot of them in this discussion.

We had a lot of questions about, you know, what kind of jobs do you see being impacted over what time horizon? So we obviously disagree to some extent on some of this stuff, but I think we answered a lot of them. I don't think disagreement is the right word, right? Because we're disagreeing about our feelings. Is that a disagreement? I guess that is a disagreement. You can disagree about feelings. I kind of think of a disagree disagreeing on facts but no you're right.

I don't think we're no I think we're all agree on. Most disagreements are about feeling. Sorry. Sorry, personal. That's a good point. Yeah, that's a good point. All right.

Okay. Disagreements is the right word then. Shades. We're all on a spectrum, right? We're just, we're falling on the AI question. We're falling along that spectrum in different places, I think. Got it.

Got it. Okay, so I like this question. It's true. That's true. It's funny. Oh, sorry, Mercer, go ahead. Okay, this is apropos, I think. Apropos. Do we consider ourselves economic optimists or pessimists, and where do you think these leanings came from? Ooh, I love that question. I love that question. Did you want to go first? Sure. Yeah.

I would say I'm an optimist. I would say I'm an optimist in general in life on my outlook. So I don't have different feelings about the economy than I think I tend to have in all aspects of what I look at in my life. I don't know why I'm an optimistic person, but I am. I think history, especially with economics, I think history has generally, and I very much hesitate to say this because I'm going to say this and the world is going to fall apart tomorrow. But I think generally, I think the history of the U.S. economy in particular has shown that the U.S. economy is incredibly resilient and has faced a lot of challenges and bumps along the way. But somehow it seems to always come out okay for most people. with a footnote that not all people, but I just have a lot of faith in sort of, I mean, in sort of the American entrepreneurial spirit and our ability to problem solve and get

through things. And that's sort of what has borne out in history. So I think I'm optimistic about the economy, just looking back on history, kind of like my AI viewpoint, right? I look back at Y2K and the internet and I see like, yes, there was a stock market bubble and we had a, we had a recession, it wiped out a lot of wealth, people lost their jobs, but the economy ended up coming back stronger after that was all said and done. We seemed to always kind of manage through it somehow. Hey, Chris, is that your characterization of Marissa that she's an optimist generally? I think so. Yeah, I agree. I agree. Absolutely. What about you, Chris? How would you answer that question?

I'm a realist. Ah. That sounds like Chris, and that's probably true, right. It does. Yeah. All right. Long-term optimistic, perhaps. I agree with Morris.

We do tend to find a solution, although it's not always the best solution in the long run, but things seem to work out somehow. Short-term, though, I am worried, right? I am pessimistic. Yeah. I mean, short-term, yeah. I'm talking about a long view, right? Whoa, whoa, what do you mean? Because of AI?

You short-term pessimist? Or policy? Between policy, right? The significant head ones. The stock market, given how concentrated our spending is on higher income consumers, right? So those are kind of front and center risks. I don't see. Right.

I don't see a lot of, even AI, I don't see, again, I guess that goes to my story here, in the very near term, I think it's still going to take a while to really pay off or start to see some dividends. So there could be some... But your baseline, the most likely scenario in the middle of the distribution is still a reasonably positive scenario for the economy, but you're saying the risks are skewed to the downside. Correct. In the near term.

That's right. Right. Right. Well, I'm... Historically, I've been very much an optimist. You know, I consider myself a macro optimist and a micro pessimist. Because I can always find things that make me nervous and upset and worried, right? But from a macro, long-run historical perspective, especially if we're talking about the U.S.

economy, the American economy, even the global economy, I think you can say this about, in broad strokes, I'm optimistic. I mean, I think we, our system, our economic system works. I mean, it's actually quite marvelous when you think about it. You know, we can solve, it feels like I feel like we can solve any problem as long as you can make money solving it. You know, and let people have it, you know, we can solve any problem. And then, you know, we get tied up in knots every so often politically, but we find a way through ultimately and, you know, come to. together. And the other thing that makes me optimistic is when things go south, and there's always

going to be things that don't stick to script and, you know, we're going to have a financial crisis or a war or some pandemic or whatever it may be. But we then come together as a nation at that point in time, and we figure things out together and we, you know, navigate through in an in about as graceful ways, I think you can. So I'm, I'm. I'm generally optimistic about things. I am nervous about the current environment, though. I am. Policy makes me nervous.

Economic policy, foreign policy, makes me nervous, more than has been historically the case. And I'll have to tell you, you can dissent it, you know, my worries about our AI. In part, I think my angst centers around the fact that I feel unmoored in terms of my ability to evaluate what the heck's going. on. This scenario analysis is the way to handle it, but it's completely unsatisfying, you know, because at the end of the day, we still don't know. And when you have that

kind of uncertainty for such a big issue with such tremendous implications, it just inherently makes you, I think, nervous and worried and somewhat more pessimistic. It's just the nature of the beast. And even on that, I feel

optimistic long run, longer run, we'll navigate through and it'll be fine and we'll be better off for it. And we need, as you said, Chris, we, the timing couldn't be better because of the, you know, concerns about labor force and demographic, the impacts of demographics. So, but, but I, in the near term, I feel, you know, more nervous, more worried, more pessimistic, but generally I would concern myself to be an optimist. Do you think I, my, my description fits, Marissa, does that, would you, would you, would you, I think you're, I think you're known to be an optimist historically.

known to be an optimist. Yeah. At times, I've been accused of being too optimistic, right? Yes. Yes. Yes. That's more, that's been more the accusation, I think, than the other side, right? And the other side, right?

But you're going down a dark road these days, Mark. I'm worried. I was just going to say I agree with you that, like, I don't know whether tomorrow we're going to wake up to, like, an apocalypse or, like, the best economy we've ever had. Oh, you? You missed the last week, and there was a really cool post on X that said basically exactly that. I don't know whether this is going to be a zombie apocalypse or, you know, this is the mother of all shorts, or this is going to be the boom of all booms, you know, that kind of thing. We don't know how this is going to play out. That's how I feel.

Yeah, like we're on the precipice of something, but I'm not sure what. I don't know what it is. Right. Yeah, exactly. Chris, what were you going to say? I want to ask you your ranking of three existential challenges, right? There's AI. Okay.

There's climate change. And there's fiscal debt situation. Well, of those three, which one are you most worried about in the long run? Most worry about it. I was going to rank order for... Okay. Okay, most worried about. I'm most worried about it.

It's not in the long run. The way I'm going to answer this is which I'm most worried about to the least worried about. And I'm most worried about AI. I'm then worried about our fiscal situation and then finally climate change. And the only reason for that rank ordering is timing. I think AI is in our face. I think the fiscal situation is not too far down. the road, you know, comes to, comes to a head, and climate change plays out over a longer period of time.

So, you know, not next year per se, but over the next decade, two decades, you know, so forth and so on. So my level of angst is related to the timing of, you know, when those things can become a real issue for the economy. And AI, again, long run, I view that as a very positive development. It's just the getting from here to there and in the most immediate future, you know, what the risks pose. How would you answer that question? So I guess just the pushback there is that, is that, so you're advising policymakers, is that the prioritization that they should be giving to policy that they are? Should they be really focused on AI right now and not worried about? Well, no, they should be focused on all of the above, don't you think?

Yes, but given that they don't have capacity. Yes, exactly. So on AI, absolutely, yes, that should be top of mind right now, focus like a laser beam. And it's not just about what do I do if the job upheaval scenario is the one that comes to pass. We lose lots of jobs and unemployment starts to rise. They also got to be thinking about the guardrails on AI. There's the other scenario, potential scenario, and this is part of one of the scenarios that we can start, the AI falls flat scenario. is that there's all these so-called negative externalities, you know, that we need to be concerned about or think about, you know, everything from cyber risk to impersonations to terrorism to, you know, just a litany of things that we need to be concerned about that, you know, AI could be a real problem, and we don't have any guardrails around that. You know, another example, I came to the four this week, this battle between Anthropic and the Defense Department, right? I mean, in very simplistic terms, the Defense Department wants to use Anthropics AI without any restrictions.

And Anthrop is saying, no, we can't do that. You know, there has to be some guardrails on the use because we're never very nervous. It can be abused and used in an improper way, and we don't want that to happen. And this sets a bad precedent going forward. So we want to have some control over the AI. So that's another example of the things we need to be worried about in the context of the negative externalities that are created by AI. So that's number one. That's, as I said, in our face.

It's happening right now. And that's what we should be focused on. We've got to be focused on all these things, but that's number one. No? Agree with that? Disagree with that? I disagree. I think.

Disagree. Yeah. I think AI certainly is front and center. But I don't know that the government or policymakers have all that much to do here. It just feels like a lot of this is out of their control, and the market's going to figure this out. I think they should stay in their lane and really be focusing on the fiscal situation. Because if we don't figure that out, that's going to make all these other problems, dwarf all these other problems. We won't have the flexibility, right, to really even develop AI policies or to deal with client.

climate change in a significant degree if our fiscal challenge has really come up and limit us. Well, I mean, AI could solve our fiscal problems if we do it right. That's the dream. That's the view. We'll get so much productivity even. That's our productivity boon scenario. That's the upside we don't talk about very much because that's the lowest probability we attach to 10%, but that's in the productivity boon scenario.

Sorry. Even in that scenario, we still have debt to GDP rising, right? That's true. It's 120% instead of 150%. But all right, who cares? It's still way out of black. Sorry, Marissa. No, I was going to say that, that it doesn't, even in that scenario, it doesn't solve

the fiscal situation. It just ameliorates it a bit, right? Still a problem. So how would you answer that question, Marisa? I was going to answer it in the same way you were. You know, deal with AI, then fiscal debt-to-GDP ratio. and then climb it. But Chris kind of, I don't know,

now I'm wondering about, you know, the government's attempt to regulate social media, for example, that's been not great. It's kind of been start and stop and not sure they've done anything real productive there. I think there are real concerns still around, not just AI, but still around social media with all these things you're talking about, right?

But I don't know, about the government's ability to effectively make policy around it. I do think they need to try. I also have a very dim view of policymakers just willingness to tackle anything that doesn't happen in the next two to four years. That's why we don't have good, right? We don't have climate policy anymore. And we never tackle the debt to GDP ratio because it's beyond the election cycle. somebody else's, it's always someone else's problem. A.I. is an immediate potential threat. And the government could also take advantage of AI, right? And so I guess now that I'm talking out loud, I agree with your ranking. AI.

Thinking out loud. Thinking out loud. I have to talk out loud for sure. Yeah. And thinking out loud. Yes. Right. Talking out loud and thinking. Yeah. So yeah, I would, I would rank it the same way you do. I'm not quite. She said she was agreeing with you kind of going over to your side, Chris, and then convince yourself, no, maybe I'll go with Mark. Thinking out loud. Thinking out loud.

I hear you. I'm just, the other point I make on this, though, the government's going to enter in only when something really goes off the rails. And my guess is the thing that's going to go off the rails first is something related AI. This close second will be our fiscal situation. The bond market loses its mind, you know, because there's too much debt or something else that triggers it. But my guess is something's going to go off the rails with AI. One of those negative externalities is going to become apparent and they're going to have to do something about it. But great question, though. Great question.

Do you want to take another one? You want to take another caller question? Yeah, yeah. Here's an interesting, it's kind of a question, but I think it's a theory. But I think it's a little bit interesting around AI. Could increased worker productivity not happen because of AI itself, but because people are afraid of losing their jobs to AI. So they kind of like kick it into high gear out of fear that their company is monitoring them and they have to be more productive. So they're just becoming more productive. So let me make sure I got that right. The question is that productivity will improve, not because of AI per se, but because of the fear AI is going to take my job,

therefore I've got to be more productive. Ooh. Isn't that interesting? That's kind of interesting. I mean, that's not a sustainable increase in productivity. But what do you think? How would you answer that question, Marissa? I mean, that could be, you know,

maybe it's what we're seeing now and over the course of the next year. That's right, because if it's coming for your job, it's coming for your job, and there's nothing that, right, that you can do about it except adapt your skills within the confines of your particular job or company. But certainly we're seeing everywhere that companies are increasingly tying things like performance, bonuses, raises, promotions to to AI usage or related outcomes or monitoring productivity more. And we could be getting productivity that's not directly because of AI, but people are just thinking outside the box in other ways and becoming more productive. So I, yeah, I think it's perfectly plausible that that's what's happening. And that could be happening over the next year or two as AI kind of is creeping over our shoulders more and more. Chris, any views on that?

I would generalize it. I agree. But I would generalize it to the strength of the labor market itself, not some. much. I think if you go back previously to other cycles, anytime you have a labor market that's particularly weak, people become nervous and yeah, they'll put in the extra hours, do the extra effort, right? They want to hang on to their jobs. And if the converse is true, maybe they're a little bit more laid back. They don't feel as much pressure. And the company or the businesses may not have as much leverage as well. So I think it's true, but I think

it's not necessarily just due to AI. I think it's just the state of the labor market itself. Yeah, I don't, it's a great question, a great thought. I don't know how to measure it, though. I mean, and also there's countervailing forces, which had been sort of on what I had been thinking, is that if companies are investing all this energy, time, money, and AI adoption, doesn't that mean that they're not devoting as much time, energy, money to other types of productivity enhancing activities. You know, what I would call business as usual kind of stuff, because I don't have the, I can't, I have to do the AI because that potentially has enormous payoff. And I'm being pressured to do that by shareholders and senior management and everybody. I need to do this.

But that just takes away from everything else that I would typically do. And that would hurt productivity growth, not to enhance it. But I don't know how to, I don't, that feels equally as likely from a theoretical perspective, right? It's an empirical question, but I don't know how to test those out empirically. There nothing comes to mind very quickly. Does that make sense, what I just said, Chris? Yeah, it does. Yeah, yeah, okay. That's a good, another really good, we're getting some really cool questions here. We got a lot. Yeah. You want, let's see one more. Yeah, I'll just say something else on that, Mark. Yeah, sure.

I think about what you just said, right, about sort of this put, the AI push might be cannibalizing other business as usual activities that could be moneymaking, productivity enhancing. I wonder about all the investment dollars pouring into AI. I mean, we know, right, I think we had a chart in the paper maybe or in the webinar that showed business investment. been spending in AI and everything else. And it was like flat for everything else. And it's off the charts for AI. So we're just not investing in a lot of other stuff that maybe we should be investing in at the expense of AI. And what if this AI doesn't pay off, right? The scenario, the falls flat scenario. We've poured a lot of investment dollars, both private and public, into AI. And it's at the expense of other things. So that's kind of something I'm thinking about too. Yeah. Yeah. Where does it

leave other industries in productive capacity and... Right. Right. Exactly. Yeah. That's a big bet. It's a big bet. Yep. Okay. Let's take one more. Oh, and I should say you brought up the webinar. I forgot to mention, in addition to the paper macroeconomic consequences of the economy, we did put on a webinar that goes over the paper in detail. And that will also be, a link to that will also be in the show note. So that'll be there for you. So, but let's take one more question and then we'll call it a podcast. Okay. Well, this kind of is in that vein of what we were just talking about. This listener wants to know, is there a point where the U.S.'s bet slash investment on AI makes the whole AI industry too big to fail? Like, is there a scenario where the government has to

bail out some of these companies because they're just so enormous. Yeah, I don't know. I don't know. It doesn't resonate with me. It's not like a major money center bank going down, you know, where it creates all the systemic risk, potentially takes out other banks and financial institutions and impairs credit broadly and takes out the economy. That was the financial crisis, the global financial crisis. I don't, I don't know. It's hard to me. for me to paint that picture. In fact, it's just the opposite.

These guys are really going at it with each other, right? It's a very aggressive form of competition that's going on. And you can feel it. Like, I think Anthropic comes out with a version of Claude. Everyone goes over to using Claude to code. And then Open Ayat comes up with a stronger version. And then within days, a couple weeks, everyone's moving over. The competition seems very, very significant, which, by the way, more broadly, makes me more optimistic about things.

Because one concern I had about AI is you had these massive companies, these hyperscalers, anthropic and Open AI and Meta and Amazon, that they would be able to shut down competition. and if you shut down competition, then innovation starts to lag, and you don't get the benefits from lower prices, and ultimately the benefits of the technology don't diffuse more broadly, and you got a problem. I was a concern, more of a concern. It's less of one now in my mind because these guys are literally going at it. It does feel like if one of them, it would be easy for one of them to stumble, kind of fall behind and get wiped out, And then it doesn't feel like that's going to be a systemic problem.

You're just the guy who wiped him out will be the new leader. So it just feels less of a concern to me. I'm less worried about that. What do you think, Chris? I'd agree with you at certainly at this stage. I guess maybe longer term. So I'm anticipating there's some shakeout at some point, right? This fierce competition can't go on forever, right?

someone eventually, well, I believe, will end up with, you know, two or three dominant, not the multitude. And then what happens, right? Then we have some concentration risk, potential monopoly, oligopoly. And, yeah, if companies become so dependent on this AI technology being delivered by just a couple of players, maybe the government would face real consequences if one of them were to go down. Right.

If you could create a scenario, I just don't think it's immediate. immediate, yeah. They're also very intertwined with each other, this whole, right, like the financing structure of all these companies. They're all purchasing from each other, borrowing from each other, trading shares for chips or, you know, I mean, they're, and we're talking about enormous companies here. So I'm a little, it's a little black boxy to me, the whole financing, and you got private credit involved that's now pouring money into these companies and that's a little not as transparent

as some other, you know, financial transactions. So I'm just a little worried that maybe something's going on in the financial shadows behind all these companies that maybe we don't quite have a great handle on. Does that mean the government will come in and bail them out? I'm not, I think probably no. Because as you said, I don't think it's, it's all these big systemically important banks that are necessarily involved. But you could imagine if this concentration continues over time and becomes even more so that we could have a financial problem resulting from this. You know what? You changed my mind. That's a great point. Really? Yeah, yeah. I mean, I'd forgotten that that is clearly a risk. Not that these companies, you know, stumble.

But if they stumble and they cause problems in the financial system, you know, because of the impacts it has, particularly on debt. I mean, if it's equity, that's one thing, you know, then it's a problem, but it's not, doesn't rise to the level of a financial system risk. But if it's debt, if these companies have taken on a lot of debt and they stumble and they start defaulting on that debt, and that may create. broader systemic problems in the financial system. That's an issue. I don't think we're there yet. I mean, I don't think the leverage is there to a degree that would suggest that. But as you say, it is a bit black boxy. We really don't know. And the direction of travel suggests that the leverage is going to continue to increase. So if it's not an issue today, it might be an issue a year from now, you know, something like that. So yeah, I think that's a good point. But it's not,

I was thinking about in the context of if these AI companies fail, but that's not the issue. It's whether they fail in all the creditors that have financed this, this massive investment, suffer losses, and then that metastasizes to the rest of the financial system and the broader economy. That's a deal. That's a big deal. Yeah, I think that's a good point. The other thing Marissa Moran reminded me of, though, is the government is now more intertwined in these companies directly, right? Intel shares and Nvidia deals and whatnot.

So I think there's some, there is more risk than I was thinking about as well, from that perspective as well, because now we have some conflicting interests. Actually, we have a separate question about Intel and the government that was asking about what happens if something happens to Intel and the U.S. government is now on the hook, right? It's kind of all part and parcel of the same thing. But yeah, we have another question specifically about Intel exposure. Right. I guess if that happens, then it just fails, and the government's like any other equity holder, it gets, that loses. I mean, the taxpayers take it on the chin, I guess.

Would they let it fail, though? Right now that's, now they have an interest. Would they, yeah. Would they double down? Yeah, something like that. You know, we've gone down a path here. You've gone down a path. Yeah. I'm not sure where it leads. Well, it's almost by definition there is going to be a failure. I mean, right? Absolutely. There must be. Right? It must be at some point.

So I guess we'll find out. The economies of scale would have to kick in there, right? Right. Well, I didn't know. I didn't think this was going to be a pure AI podcast, but it turned out to be a pure AI podcast. And we still now have 40 questions to answer. 39.

That's getting the quad. Yeah, but these are really very thoughtful. and helpful questions. So please, please, please keep them coming. How do people send us questions? Is it Inside Economics at Moody's dot com? Is that the best way? Yes.

Yeah, very good. Okay. Well, I got to make my way to Tampa. Wish me luck. Good luck. Okay. Anything else before we call it a podcast? Marissa, Chris?

No. No? Okay. All right. Well, with that, dear listener, I hope you enjoyed the conversation, and we will talk to you next week. Take care now.