Isabella Loaiza: Beyond Exposure

Ben: Is AI going to cause a white-collar bloodbath? Or is that narrative itself part of the problem? And does the way we talk about technology shape the technology that ends up being built? My guest today is Isabella Loaiza, a researcher at MIT whose work challenges some of the most common assumptions about how AI will affect workers and jobs. Here's our talk.

Ben: Welcome to The Economics of Work. I'm excited to welcome Isabella Loaiza. Did I get that right?

Isabella: Yes, that's correct.

Ben: Nice. Isabella has been influential in changing the debate about how AI is affecting work, and in adding some very important nuance to it. Isabella, welcome to the show.

Isabella: Thank you so much for having me, Ben. I'm so excited to be here.

00:38 What we get wrong about AI and work

Ben: Let's break down this debate about how AI is changing work, and what you don't like about it. Feel free to throw some stones. What are we getting wrong in how we talk about AI and jobs, AI and work?

Isabella: Oh, wow, so many things to discuss. To begin, I'll choose two things I often think about. One is the narrative that AI is going to replace all workers, that it's going to be a white-collar bloodbath kind of situation. I think we're getting that mostly wrong, for two reasons. The first is that technology will do what we want it to do. If we build for replacement, that's what we're going to get. If we build for augmentation, we'll probably get some augmentation. And yes, some roles will definitely be replaced, because that happens with all technological change. But how we design the technology is going to have a very large impact on the effect we actually see in the labor force.

01:39 Who controls what gets built?

Ben: Let me drill into both of those points, and I promise we'll circle back. This idea of a white-collar bloodbath, that's a prediction.

Isabella: Yes.

Ben: Which I also think is probably wrong, but I'd love to find out where the chain of reasoning breaks down. And on the other point, I want to note it so we can circle back: this idea that we're in control of what we build and how the technology we build will affect jobs. That sounds hard. Technology isn't organized by some central planner. A lot of forces determine how it evolves, and I wonder whether that's in anyone's control. You can answer those or circle back, whatever you want.

Isabella: I agree with you. Technology is not centrally designed, and we're happy that's not the case, because we want innovation everywhere. We want people to be able to create and build the things they think are necessary. I'm a big fan of that, for sure. But from my position at MIT, and having gotten my PhD at the Media Lab, I work very closely with people who actually build the technology. Because I'm in an environment where a lot of the folks creating the ideas and thinking about applications are right there, I can see that they do have some control over how they frame their applications, their innovations, the new products they're trying to build.

We can't control all the companies out there. But just as sci-fi narratives shape what we think is possible, what technology designers and creators think is possible is going to shape the applications they think of when they're building new products. So yes, technology is not centrally planned. But by helping build a more augmentation-centric, human-centric narrative, we can inspire the folks who build the technology to build it in a way that doesn't just replace, but also augments.

Ben: Fair enough. So as economists and thinkers about this, we also need to be inspirers of technology builders. It's a new role for us.

Isabella: For sure, I completely agree. And it's not just the technology builders. As people in organizations start to deploy these technologies, they'll want to get creative and think beyond replacement: How can I enable my workers to do things they couldn't do before? What can I do as a business? Is there a product or something I've wanted to do that I haven't been able to, because there was no technology for it? The way we implement the technology will also have a huge impact on whether AI replaces or augments workers.

04:16 Automation vs. augmentation

Ben: So this idea of automation or augmentation: is that the right spectrum to think about? In your mind, is the spectrum between automation and augmentation the same as the spectrum between substitution and complementarity, or are those different concepts?

Isabella: That's interesting. Earlier this year I was thinking a lot about this, because some of my colleagues consider automation and augmentation to be different flavors along the same continuum. Some folks think augmentation is partial automation. I'd push back on that a little, because I think there are two parts to augmentation. One part is the partial automation story. For example, I no longer have to write my emails from scratch. I can say, "Hey, I want to say roughly this," and the AI writes the email for me, so I can be more productive. That's partial automation.

But then there's the augmentation that comes from being able to do things I couldn't do before. That has nothing to do with automation. It has more to do with enhancing my capabilities: being able to think of things I couldn't think of before, or produce things I couldn't produce before. And that's the part of augmentation we usually miss when we try to measure the impact of AI on the workforce right now.

Ben: I think that makes sense conceptually. But you could also make the case that with partial automation, where you automate some subtasks and focus more on the remaining ones, or introduce new tasks, the bundle gets augmented, and that can also be enhancing. It doesn't necessarily mean less work. It could mean you have more time to do bigger and better things, and that we can build things we've never built before. Pure augmentation is like having shoes that make you run faster, or riding a bicycle: something that purely levels up your capabilities.

Isabella: You can even take it to the extreme. Instead of shoes that make you run faster, maybe shoes that make you fly. That would be actual augmentation.

Ben: Exactly. You enhance your own capabilities.

06:37 Pure augmentation or offloading tasks?

Ben: But I wonder if that distinction is so useful in this conversation. I'm a little torn on this, but when I think about AI, I think of it as primarily executing subtasks that are part of a broader workflow, not necessarily doing things that propel us forward. Let me put it another way. I don't think of AI as shoes that make us fly, or a bicycle for the mind. I think of it as letting us offload parts of what we do, and maybe getting better performance on those subtasks. That complements our remaining work, which can be augmentative at the job level. But do you think there's some pure augmentation effect from AI in particular?

Isabella: I think the augmentation you're describing is a kind of spillover effect. We remove some tasks from our bundle, as you called it, and then we get the spillover: more time to do other tasks, or tasks we didn't do before. I completely agree with that, and it's a distinction I hadn't included in my earlier conceptualization of augmentation as partial automation. I do think we can get some of that from AI. AI can take over some tasks we already do, and then we get that spillover.

One thing that might be interesting to note, though: in principle AI can take over some of these tasks, but I wonder how far we get this spillover augmentation if the tasks that remain are so cognitively demanding that knowledge workers burn out more, which we're starting to see surface in the literature. And the tasks AI does take over can end up making us spend more time on them, which is the idea of workslop that's also appearing in the literature. What do you think about that?

08:39 Can losing easy tasks cause burnout?

Ben: I don't know. I haven't thought much about this burnout effect, so let me make sure I understand the mechanism. Say you have AI doing a bunch of stuff, and you have to specialize in the bottlenecks, and those bottlenecks are just more mentally taxing. Do they just happen to be more mentally taxing, or are they burnout-inducing by construction?

Isabella: From what I've seen in the literature, the story goes something like this. Knowledge workers have cognitively demanding tasks. An example from my own work: I have to come up with interesting research ideas. I need to think about how to communicate them to other scientists and to the general public. I need to make beautiful, stunning visualizations. These are all cognitively heavy tasks, because they require creativity, critical thinking, and judgment.

But I also have smaller tasks, like emails and administrative work. What usually happens throughout my day is that I spend some time on the highly demanding cognitive tasks, and when I want a bit of a rest but still want to be productive, I shift over to admin and email. Then, once my mind is rested, I go back to the demanding work. So if those smaller tasks no longer exist, which is great because I don't have to spend time on them, I'm only doing the highly demanding cognitive work and I get no rest. That leads to higher, or faster, burnout in knowledge workers.

10:12 Why not just take a coffee break?

Ben: That seems like a very nice problem to have, in a way. Our time is so valuable, and we're doing such critical linchpin tasks, that if we spend all day on them, we're insanely productive. We're being used for what we're most productive at all the time. And instead of relaxing by doing menial tasks, couldn't we relax by having a cup of coffee? Couldn't we just do nothing? Wouldn't that be at least as relaxing as doing the automatable tasks?

Isabella: It would certainly be ideal if, instead of relaxing by checking my email, I went and had coffee with a friend. That should be feasible, but there are a couple of things at play. First, some workers are in different conditions. If they feel they need to stay at their desk to show they're working, they can't go grab a coffee with a friend. That's a very specific context. The other is challenging a very entrenched narrative about what it looks like to be productive at work. Especially here in North America, a lot of workers take real pride in being very productive, and going to get coffee doesn't feel as productive as checking your email, even if it's the best thing you could do to rest your mind so you're more clearheaded for the highly cognitive work. So there are several things at play, as there always are with work and the workforce.

11:48 What AI exposure gets wrong

Ben: I want to talk about AI exposure and your framework, EPOCH, how those differ, and the best case for the framework you created with Roberto Rigobon. Super fascinating work, by the way. I'd love to get your thoughts on where it improves our understanding of things.

Isabella: Of course. The framework we built is called EPOCH, and the idea started because a lot of the literature I was seeing had a doom-and-gloom narrative to it. Roberto and I thought there had to be a better way. Technology doesn't just substitute for human labor; it also complements it. And that isn't just a purely optimistic story. It's something we've seen happen historically. So given that folks are so focused on what machines are going to "take away" and replace, let's also think about what humans can still do that's valuable and that complements what AI does, so we can have more productive workers. That's where it started.

Something else I saw in the literature that I feel quite passionately about is the term "exposure." I think exposure is perhaps not the best metric out there, because it bundles two effects that can push in opposite directions in the labor force: automation and augmentation. Bundling those together can give us conflicting insights. For example, I think the latest paper by Morgan Frank and colleagues found that workers in highly exposed jobs spend less time looking for a job in the current labor market. That's counterintuitive, but it's the augmentation part at play instead of the substitution part. By separating the two effects, we get a clearer picture of what we're measuring and what impact AI is having on the labor force.

14:01 In defense of exposure

Ben: Let me offer a defense of exposure, which I know is kind of a punching bag. I agree with you when exposure is analyzed at the job level. But if exposure is analyzed at the task level, it's really just automation. There's no real augmentation of subtasks.

Isabella: I agree.

Ben: The task is just being automated. How that gets aggregated up to jobs is where it gets really ambiguous. If you say 50% of a job is exposed, that may substitute for some work, or it may complement some of the remaining work. That's the big question. And it sounds like that's your main critique: the metric once it's aggregated to the job level, where the aggregation function is totally unknown.

Isabella: Yes.

Ben: Maybe unknowable. But that's our job, to understand it. I don't know if anyone has really tackled it.

Isabella: For sure. And I completely agree that at the task level there's no real augmentation; it's just replacement. But if it's just replacement at the task level, then don't measure exposure, measure replacement directly, so we have a clear idea of what we're measuring. Part of my qualm with it, and Roberto's too, is that we're obsessed with measuring things correctly. Even if you think a metric is a good approximation of something, when you apply it across the 19,000 tasks in O*NET, or however many jobs you're looking at, the measurement error might paint a different picture than what you think you're measuring. So if you want to measure automation at the task level, by all means, be my guest. We do that in our framework too. But don't call it exposure. That's the first point.

Ben: I'd interpret exposure as something like automation potential, not actual automation, because to automate you also need adoption, and that happens through a different process.

Isabella: For sure.

Ben: But you'd feel better if it were just called automation potential.

Isabella: Yeah, I would feel better if we called it that. The other thing I feel a little iffy about is the way it shows up in news headlines and the way folks frame it in their research papers. Exposure is mostly interpreted as automation. But when you push a little further and say, "Well, AI doesn't just automate, it also augments," they say, "Oh, yeah, but exposure measures both." You can't have it both ways. You can't frame it as automation, and then, when people point out complementarity, retreat and say it also encompasses that. You need to keep the two separate. That's my thing. It might be a little nitpicky, but as an academic, I feel like it's my job to be nitpicky.

17:06 What EPOCH stands for

Ben: No, this is the nuance we need. I'm sold. So, EPOCH in particular. It's an acronym: Empathy, Presence, Opinion, Creativity, Hope.

Isabella: Yes.

Ben: Very cool acronym. I think of these as the things that are uniquely human, but is that the right way to think about it? Would this be the same set of things for any technology, or is it specifically the things AI in particular can't do?

Isabella: We designed the acronym for AI specifically, both traditional AI and generative AI. We wanted to capture the things that, at the current moment, are still very much in the realm of human knowledge, where humans are uniquely positioned. Here my co-author and I diverge a little. Roberto very passionately defends EPOCH as things that will forever remain the realm of human expertise. I'd be a bit more careful. I think some of these things will continue to be human expertise now and in the near future. But I try to be very careful about speculating far into the future, because I don't know what it will look like. I don't think we ever imagined things like LLMs and all the technologies we have today. Human creativity and ingenuity are so wonderful that we don't know what we'll have in the future. So I want to be careful and say that, at least right now, these are the capabilities that remain in the realm of human expertise in the AI economy.

18:46 Which human skills AI could take next

Ben: If you had to guess, which of those elements is most likely to drop off as technology gets better and better?

Isabella: Interesting. Since each letter is a bin of capabilities, I'd think of specific things that could drop off rather than an entire bin. For example, right now AI isn't very good at causal reasoning. I don't see why we couldn't build a technology with a different architecture that could do causal reasoning as well as humans.

Ben: Which bin is causal reasoning in?

Isabella: It's in Opinion, bundled together with critical thinking and judgment. I still think humans will continue to be critical thinkers, and will need to be, now and in the future. But causal reasoning isn't a biological constraint. We just haven't found a way to make machines reason in a way that's sound with human logic, rather than replicating patterns in the data. I think that can be done in the future. That's one.

Another that could be achieved is systems thinking: being more aware of context and seeing how different things connect in a bigger picture. That's in the Hope bin, along with leadership and motivation. I think that one's achievable too.

And one that's quite contentious is empathy. We're seeing so many studies saying you can use AI for therapy, and that AI can even be more empathetic than humans, because it doesn't get tired of hearing you complain or of you asking the same question over and over. I feel a little conflicted about that, because it's simulated empathy. It's a different kind of empathy from human empathy. I don't know what we'll call it in the future, but some people would argue that empathy is very nearly automatable. I'm a humanist at heart, so I'd say it's a different kind of empathy, and I hope it doesn't get completely automated away. But it's an open question.

21:06 Can a machine be empathetic?

Ben: I'd think of empathy as, on the one hand, a capability, but also as a bit more demand-driven, more of a reaction. Someone is empathetic if someone else interprets them as empathetic. Say you took your favorite musician and found out they were actually just an AI. I wouldn't really listen to them anymore, because what I like about my favorite musicians is that I think they understand what I'm going through. They've been there.

Isabella: Yeah.

Ben: I can relate to them, and they can relate to me in some way. I just wouldn't feel that if it were simulated.

Isabella: I completely agree, and there's research that supports what you're saying. I can't recall the specifics right now, but they did an experiment where they presented people with empathetic statements, written either by AI or by humans, and people chose the ones they liked most. In the end, a little more than half chose the AI statements as more empathetic.

Ben: Oh, wow. Okay.

Isabella: But then, when it was revealed which statements were from AI, people still preferred human empathy, even when they'd rated those statements as less empathetic. That's exactly what you're talking about. We tend to prefer human empathy to synthetic empathy because it's about human connection. It's not about the output or how it looks. It's about the connection between two humans.

Ben: Interesting. I remember seeing that, and I may not get the specifics right, but I remember that people didn't feel that way about doctors in particular. The theory was that people very often feel their doctors come in like robots, "here's this and that," and bad doctors, doctors without bedside manner, almost have negative empathy value. When an AI communicated instead, people actually reacted better. That makes me hopeful there will be increasing returns to humanity for various professionals.

Isabella: Oh, for sure. I have no doubt that AI is going to create a lot of positive externalities for some people. The goal is to make those externalities reach as many people as we can, not just a select few.

23:42 Creativity and what EPOCH leaves out

Ben: The element of EPOCH I'm a little more suspicious about is creativity. Sometimes I see what AI produces and think, "Eh, whatever, I'm not that into it." But sometimes I think, "Wow, you really got somewhere I didn't, and that takes some creativity." It's very impressive. Who knows? I just want to register my prediction in case we revisit this in 10 years.

Isabella: Lovely.

Ben: And see what AI can do. Since writing the paper, is there anything not on the list that you now see as uniquely human and would want to add?

Isabella: Interesting. Not so far. When I teach this, I always like to ask people what the H stands for, because I think that's the most surprising bin in our acronym, and I like to hear what people say. Quite often they say humor, or humanity.

Humor I find quite interesting. With machines, sometimes their humor lands and sometimes it doesn't. They can do very basic jokes, but the more layered things actual comedians do are still very out of reach. And with humanity, I wonder what people mean, because humanity can be so many things. It's such a broad concept. But it comes up again and again in class.

25:16 Would an EPOCH workforce be a good world?

Ben: My first instinct is that it's all humanity, and that's the umbrella term. But I'd think of humanity as more closely aligned with empathy. The H is actually my favorite one. I really love the idea that hope and motivation are things we uniquely add, and that they're a source of value.

What's really absent from this framework is intelligence. Not everywhere; like you said, it's in causal reasoning and systems thinking. But if we take the collection of capabilities of working humans today and compare it to the EPOCH collection, the EPOCH collection is less intelligent. And I wonder whether a world where everyone becomes an EPOCH worker is a good world. On the side of it being a good world, it feels very human, and the whole vibe of it feels very nice. But I really like intellectualizing things and being analytical. For me at least, that's a part of work I enjoy. So in your opinion, would you welcome an EPOCH workforce, or would you pine for the past a little?

Isabella: I need to make a couple of distinctions. Something we tried to emphasize in the paper is that to be an EPOCH worker, you don't need all the capabilities we mention. You just need a really high requirement in one of the bins to have a high EPOCH score.

Ben: I'm really thinking in aggregate: the workforce as a whole specializing in this collection of areas where we find unique value.

Isabella: I think that's an open question. Of course, I'd think it's a better world, a "good" world, though I don't know what a good world is. I'm so sorry, I'm so nitpicky about terms; the academic in me can't help herself. But if people enjoy doing these kinds of tasks, then for sure it would be a better world if we enjoyed our work more than we do now and were doing the interesting kind of work.

I'll push back a little on intelligence, though. I agree we don't have the specific term "intelligence," but Opinion and judgment cover a lot of what you described enjoying about your job: intellectualizing things, analyzing things. That's in Opinion, along with critical thinking. So you could be a high-EPOCH worker and still do all the things you enjoy about your work.

It will also come down to what people enjoy about their work. Lately I've been thinking about running an experiment on this. People enjoy some tasks more than others, so how do they report enjoying them, or how motivated do they feel, if you tell them a task is more automated than the others? I haven't run it yet; I'm still figuring out the right setup. But I think it'll be very interesting. What happens if the part you like most about your work is automatable? How do you feel about that? I don't know. That's why I'm trying to research it.

28:33 What leisure reveals about good jobs

Ben: If you go down that road, one thing I'd love to see is what capabilities people want in their friends. Or set work aside entirely and look at what people do in their free time. We need some capabilities to be good at leisure: to travel, play sports, socialize, make dinner, whatever it is. That would be a kind of revealed preference, because if we're exercising those capabilities in our leisure time, those are surely things we enjoy, or we'd have chosen something different. I wonder if that's a way to get at what makes a good job, or a fulfilling job.

29:22 Dream jobs vs. meaningful jobs

Isabella: Oh my gosh, you just said "fulfilling job." This is one of the things I really want to research in the future, because I've been thinking about dream jobs and meaningful jobs. I don't think the two are necessarily the same. They can be very different, and different in different contexts. Your dream job might be working in finance, but maybe you dream of that job because it's going to bring you a lot of money, not because you necessarily enjoy the work.

Ben: Sure. Chasing riches.

Isabella: Exactly. A meaningful job might be, "I want to be a nurse because I want to help people," even though it may not pay as much. They can be very different things, or they can coincide along some axis for some people in some places. I've been thinking about this question so much, and I really want to research whether what's considered a dream job or a meaningful job changes in the AI economy. We'll certainly see some historical changes that aren't attributable to AI, but the question is whether what you find meaningful changes because of AI. If AI can do something as well as a human, is that still meaningful to do as a job? I wonder about that often.

30:35 Why active fund managers still exist

Ben: That makes me think of another line of your research that I found really interesting, on technological adoption in finance. With apologies to any mutual fund managers out there, I think of theirs as a job that's been fully automatable for a very long time. There's research going back decades showing that active managers underperform passive indices, and yet there's still demand for active fund managers. I wonder if that's meaningful work in the same way, or if the people who do it have to convince themselves they're providing value. How do you see the relationship between these limits to adoption, in fields like active fund management, and the idea of people finding their work fulfilling?

Isabella: That's interesting. If I were thinking of meaningful or fulfilling jobs, I probably wouldn't have gone to mutual fund managers. I'm sure they find it fulfilling in some ways, and rewarding monetarily for sure, but I don't know if the meaning mechanism is what's at play there. When I think about whether the technical feasibility of automation is the same as the adoption or automation we actually see in the market, I think of tax preparers. We've had TurboTax and similar tools for many, many years, and we still have tax preparers. There's still demand. I assume they handle the more complex cases, not the very simple return I file every year.

When it comes to meaningful jobs, I'd lean toward things that are creative or empathetic: jobs in the healthcare sector, the creative sector, social work. Those are the jobs people seek out for being fulfilling. Or academia: theoretically, you become a professor because you want to do research and teach. So in academia, even if AI could replace a professor, I'd still expect demand, both from people who want the job and from people who want to be taught by a human professor rather than a robo-professor. For fund managers, I'm not sure how the meaning dimension plays in.

33:07 Meaning and being uniquely useful

Ben: I've got to imagine that how automatable it is factors in. A personal anecdote about teaching: I used to teach Python and data science at NYU's business school, and I started finding it less meaningful, mostly because there were so many good alternatives. You could learn Python and data science from MOOCs, Coursera, and all these really, really good resources. I thought, "I'm not really doing a better job than these technological alternatives." Then I started teaching the future of work, and I love it, because I feel uniquely able to teach it. I'm adding something that isn't already free on the internet. That's a source of meaning for me: feeling I have some unique capability to produce this.

Isabella: So you're attaching meaning to how uniquely positioned you are to do it. Not "I want to help people," but how unique you are in doing it.

Ben: I think so. What do you think?

Isabella: That can definitely be a source of meaning, but it's not the only one. Personally, I'd align with that in how I find meaning in my work. I came to academia because I wanted to help the world, and I thought that with my particular set of skills, I was more uniquely positioned to help in academia than elsewhere. But other folks don't think about their competitive advantage as part of what makes their job meaningful. Some people just like interacting with others and find helping meaningful. I don't know if they compare how well they can help here versus there. Maybe it's just an inherent preference. I think that's interesting.

And on a different note, about your teaching Python and data science: I agree there are so many amazing online alternatives, but there's also research showing that people often don't finish these courses, and that MOOCs haven't been as transformative as we expected at the beginning. Some of that might be a lack of motivation to finish. And I think Khan Academy just scrapped Khanmigo, their automated tutor for kids, because they found it wasn't as helpful as a human tutor.

35:38 What Isabella wants to change

Ben: I think that's right. There's tremendous value in being a motivator and a tutor as a teacher. Maybe mine was really just personal: I was mostly interested in being an expert as a teacher. That was a preference, I guess.

You mentioned your motivation for public impact, which I love. It's very exciting. What do you think can be done that isn't happening yet? What are you excited about? How can we, the broader thinkers in this space, change meaningful outcomes for people?

Isabella: I see a couple of avenues. The first, which I'm very excited about, is shifting the narrative to look at the opportunities AI brings, not just its use as a replacement tool. There's still a lot of work to do there. Maybe the narrative in the US is changing a bit in 2026, but in other regions of the world it's a message that still needs to be discussed. So that's one thing I'm very excited about.

Another thing I'd really like to do, which I don't see people in my field, the future of work, thinking about: a lot of the work focuses exclusively on how AI can make us more productive. But AI can do so many other things for us. It could potentially help us collaborate better with others, and I don't think that's been explored much. Right now, a lot of the research is based on the individual worker, not on the worker and their team, or the worker and the organization. That's a huge area I'd be very excited to think about.

And the other area, which might be a little more hippie for my colleagues, is what we were just talking about: how AI can help us find more meaningful jobs, or dream jobs. How can we use AI to build dream jobs and meaningful jobs that make workers happier and also more productive? That's my hippie side showing a little.

37:50 Why poorer countries fear AI more

Ben: That sounds exciting. You mentioned that in the less developed world, and let me know if I'm misunderstanding, there's more pessimism about what AI will bring. Is that right? Why do you think that is?

Isabella: I don't have the statistic in front of me, but-

Ben: You think the vibes are different?

Isabella: Sure. Especially in some countries, when we say AI is replacing routine white-collar jobs, those countries have specialized in exactly those jobs. So they're asking, "Oh my gosh, if this has been my winning horse for so long and now AI is going to take it away, what's going to happen to my economy?" Perhaps we need to help them think of ways to shift, or ways to augment their workers, so they're not just suffering the effects of replacement but can also develop economically. I think that might be contributing to the narrative of fear.

The other thing is access and familiarity. Here in the US, everybody's talking about AI, and most people I interact with on a daily basis have used it for some task. In other contexts, AI is a very abstract idea: we don't know how it works. As humans, we tend to be afraid of the unknown, so helping people become more familiar with AI can also be a way to dispel those fears.

39:18 Measuring work around the world

Ben: On that note, one final question for you. Say you were in charge of creating a global labor market measurement system. What would you do?

Isabella: Oh, wow. A global labor market measurement system.

Ben: You could take it in whatever direction you want. Something like an O*NET for the world. If you were running the global department of labor and had to put out some data, what do you think would be important?

Isabella: I would measure so many things. And I'd really try to emphasize regional preferences, because there definitely are regional preferences.

Ben: For work, or for goods? What do you mean by preferences and work dimensions?

Isabella: At work. Something I often think about: I'm originally from Colombia, and I think there's a very large difference in what workers are expected to do in the same roles. For example, an accountant here in the US may require more technological skills than one in my hometown. But in my hometown, the accountant might need more social skills, because Colombia is a high-context culture, and you need a lot more coordination and communication to accomplish your tasks. Those are huge differences that O*NET doesn't capture. I'd love to measure how the content of work changes across regions, in terms of what it takes to actually get work done. Right now, our taxonomies are really hard to get to talk to each other, and they don't cover all the different tasks people have to do across regions. So that's something I'd be very interested in measuring.

Ben: Cool, that sounds exciting. I hope someday that becomes your responsibility. I think we'll have a better world for it. Thank you so much for being part of the show.

Isabella: Thank you so much for having me, Ben. This was very exciting.

Ben: The Economics of Work is brought to you by Revelio Labs, workforce data for research and benchmarking. If you enjoyed this episode, please remember to rate, review, and subscribe. It'll help us reach other curious listeners. I'm your host, Ben Zweig. Our producer is Cole Wagner. Thanks for listening.