Afford Anything - AI, Layoffs, and the Future of Your Career — with Dr. Ben Zweig (Part 1 of 2)

Episode Date: February 27, 2026

#693: AI learns your job in weeks … and you start wondering if you still have one. That question shapes our conversation with Dr. Ben Zweig, CEO of Revelio Labs, a workforce data company that uses AI to build large employment databases and study labor market shifts. He also teaches a class on The Future of Work at NYU Stern School of Business. He holds a PhD in economics from CUNY Graduate Center. Dr. Zweig starts with the legend of John Henry, the steel driver who raced a steam drill and lost his life trying to prove that a human could still beat a machine. The story mirrors the Luddites, who smashed looms when automation threatened their work. The fear of technology replacing workers is a theme throughout history. It keeps repeating. And yet, this time it feels different. You hear how today’s panic fits into a longer pattern. Sixty percent of current jobs did not exist a century ago. Even jobs that kept the same name changed completely. Dr. Zweig describes his father tabulating punch cards as a statistician, while he now builds neural networks. Same field. Different tasks. We break down what a job actually means. A job is a bundle of tasks. You execute tasks, but you also orchestrate them – deciding order, workflow and coordination. AI tends to automate the most granular tasks first. Broader, abstract orchestration proves harder to replace. Dr. Zweig argues that “augmentation” often just means partial automation that frees you to focus on what remains. The discussion turns to empathy-driven roles, such as rabbis, psychologists, and teachers. Dr. Zweig cites traits such as empathy, presence, opinion, creativity and hope as distinctly human. He notes AI still struggles with memory and long-term relational trust. You also hear what this means if you are early in your career. Hiring has slowed. Entry-level roles appear more exposed to automation. Dr. Zweig says younger workers often lack orchestration experience and face a risk-averse market. He says that to be competitive in today’s job market, you should take ownership of complex projects from start to finish. Show people – through networking and demonstrated work – that you can manage more than just tasks . Learn more about your ad choices. Visit podcastchoices.com/adchoices

So if you haven't seen the headlines yet, there's this company called Block. It's the company behind Square and Cash App and After Pay. It just announced that it's laying off 4,000 employees. That's roughly 40% of its workforce. It's going to go from over 10,000 employees down to under 6,000. With this news comes the question, is this a harbinger of things to come? Will we all be unemployed? And that question itself is a subset of a much bigger question, which is, What is the future of work? To answer that, we brought in the guy who teaches a class called The Future of Work at NYU Stern School of Business. His name is Dr. Ben Zwegg, and he's the CEO of Ravello Labs, which is a workforce data company that uses AI to build big employment databases and track shifts in the labor market in real time. So his company analyzes millions of job postings

and analyzes employee records to see. what's actually changing in the world of hiring. Dr. Zwegg holds a PhD in economics from the CUNY Graduate Center and he is the author of Job Architecture, which is a book that traces how we structured work in the first place from the founders of Wall Street to early management consultants to modern data scientists who are trying to make sense of the labor market. So he, in that book, he argues that many of the systems around how we work were both for a different era, and AI is really exposing that. And of course, that's going to influence the future of work. So today's episode is part one of two. In today's episode, we are going to

talk about which jobs are most vulnerable to automation. We're going to talk about how, in particular, this affects younger workers, why hiring has really slowed for younger workers. And we're going to discuss what skills matter in the era in which AI can execute. you many, many tasks faster than humans can. We'll talk about what parts of your job AI can already do and what that means for you, particularly if you're early in your career. That's all in today's episode. And then in the next episode, part two, we will go deeper into the taxonomy of work and how all of that is going to likely get restructured as we move into the next few years. Oh, I should introduce myself. Welcome to the Afford Anything podcast, the show that knows you can afford

anything, not everything. This show covers five pillars. Financial Psychology, increasing your income, investing, real estate and entrepreneurship. It's double eye fire. I'm your host, Paula Pant. And today's episode, you know, we talk about the five pillars that the show covers. Today's episode is about that first letter I, increasing your income, because, well, any discussion around jobs, careers, around the future of work traces to that. How do we continue to earn more to thrive in our careers in a landscape that is changing so rapidly. Again, today's episode is part one of two. Today's episode is shorter than what we usually publish, so we're going to do something

a little unusual. We're going to take just a couple of minutes to hear from the sponsors who make this all possible. Today's episode is a shorter one, so I didn't want to interrupt our conversation twice. That's why we're leading with this word from our sponsors first. That way we can minimize interruptions when the interview actually begins. and so we're going to hear from them. And after that, we're going to go into part one of this two-part series with Dr. Ben Swag.

Welcome. Thank you for joining us. Yeah, thanks for having me. Happy to be here. Can you tell us the story? I think a lot of us learn this as kids in elementary school, but I'd long forgotten it. The fable of John Henry. John Henry is really an illustration of the Luddites.

The story of the Luddites are these workers who were weavers. And then the loom came out, which is this machine that can do weave them. super fast, you know, 100 times faster than a human could. They smashed the machines. This story of John Henry is the apocryphal story of that. This is such an interesting history because for so long, the Luddites were considered these bozos that just like could not get a handle on technology. They were so against technology automating their jobs.

And recently, there's been a kind of trend to kind of reclaim the history of the Luddites. and say maybe they weren't such bozos. Maybe they were really actually quite rational. And we're right to be kind of fearful of losing our jobs to automation. I feel like that's maybe gone a little too far. But yeah, I mean, this story kind of keeps coming up again and again in different versions. And we have the same idea now. We have union workers going on strike because they don't want to be replaced by automation and things like that.

So even though it gets told in like lots of different ways, it touches on this real fear of losing our jobs to machines. And how do we process that fear right now? Because we can look back at examples in history where certain technologies have come obsolete, but new technologies have arisen. You have a stat, something like 60% of today's jobs did not exist back in 1920 or 30 or 40. Yeah. Yeah.

And I think that really understates it because those are new jobs. even the composition of jobs change entirely. So even jobs that do exist from period to period, we do totally different things in them. A personal story, my dad was actually a statistician, and I'm kind of a statistician. You know, I'm an economist, and I analyze data, and I consider myself not a statistician.

I think what he was doing in the 60s 70s was tabulating frequencies on punch cards to detect some anomalies and look for fraud and stuff like that. And what I'm doing is like building neural networks to create embeddings of job titles to find out what job titles are similar. And that is so different, even though the occupation that we're in might be fundamentally the same. Which actually then brings me to the question. I guess we should start with by establishing the definition of what we're talking about. What is a job?

Because you differentiate a job from the microtasks. And that also is differentiated from the orchestrator. Yeah, okay. So, yeah, let's go over this vocabulary. What is a job? I think the almost perfect definition of a job is a bundle of tasks. Mm-hmm. It's not quite the right definition,

but I think it's a very useful step forward from thinking of jobs as the atomical unit of analysis. Sometimes people talk about automation as affecting jobs, but jobs are combinations of things. So I think it's automatically when we talk about automation, we really can't be talking about jobs because we can go more granular. than jobs. And we can think of jobs as a collection of things. Your job as a podcaster or podcaster plus, I guess, is probably a collection of different things. You're not just doing the primary activity. You're also preparing and analyzing social media. Like, there's a whole collection of things.

And any person who has a job, they have a dozen or so responsibilities that they do in their job. The job title is really kind of a shorthand for a collection of things. So we know when someone says they're a teacher, they're doing some instruction, they're doing lesson prep, they're doing grading, they're doing a whole set of things related to teaching. And each of those tasks can be broken down into finer, kind of more detailed sub-tasks.

That is like a way to think about what a job is. To get even a little bit more precise, when we do a job, we are not simply executing on the tasks in the job. We are also orchestrating those tasks. So we're also deciding what should come first, what should come second, who depends on this, how do I combine these tasks into a workflow to minimize the amount of work, maximize the output, who do I need to work with, what are the connections that need to be made?

So we are both executor of tasks and orchestrator of tasks. And I think that's a useful framing in a few ways, because one is that when we think about technology, is technology doing more of the execution? Is it doing more of the orchestration? So how do we complement that? That's like a big question that I think we need to grapple with. Another question is, why do jobs exist?

There's some discussion, there has been some discussion. There's been a trend a few years ago about maybe we should just unbundle jobs entirely and just transact and tasks. So there was this wonderful book written called Work Without Jobs by John Boudreau and Robin Jasutthuson.

Really fun book. basically makes this argument for unbundling work and just doing the work that needs to get done. And what do you need job titles for? I don't fully agree with that, but I think it's a really useful way to think about work because at the end of the day, the organization really cares about the work being done, the delivery of the output. And having that be delivered through a job is a second order priority, maybe not important at all.

Maybe it could get done through a freelancer who's like delivering on a specific task. Maybe it could be done through a technology or a vendor. So it's useful to think about the delivery of the work activities of the tasks as distinct from the jobs and the people who hold those jobs and the skills that they have. So I think it's useful. But thinking through this coordination also forces us to think through, you know, what are the costs of coordinating these different work tasks? So if you've got like a collection of things that you do as a person,

it's not that hard to coordinate. Like we can orchestrate ourselves pretty seamlessly, much more seamlessly than what it would take to coordinate a set of activities between people. So if you're working on a team, there's more friction than just like coordinating your own activities. And if you're working between teams, there's even more friction. And between companies, there's even more friction. So we have this tradeoff between core. coordination costs, which are kind of low at the micro level and higher as you get more bigger

and more abstract, with specialization, which we get more of when we have bigger entities. These are just important dynamics to think of as we think through why does work look the way it looks and how will that look in the future. Right. When we think about the execution of jobs versus the orchestration of jobs, it seems to me, and please correct me if I'm wrong, that right now, these early AI models, they can execute on specific tasks, but are they currently able to orchestrate workflows and orchestrate how a bundle of tasks are performed? And if the answer is no, then are they moving in that direction or will humans always be needed to orchestrate? Yeah, it's a really great question. It's a huge question. So I'm only kind of take you on this

journey where I kind of changed my mind about this a little bit. At my company at Rebellio Labs, we have this taxonomy of work activities. And for every activity, we want to know how vulnerable that activity is to automation. And this taxonomy of work activities is hierarchical. So you have these broad levels,

which could be like analyzing financial data. And then you have like really granular levels, which is like implementing Excel macros or something, like things that are much more detailed. You could think of a collection of interrelated tasks as a workflow. but what we found is that if we ask a generative AI model to score each task on the risk of it being automated,

the ones at the very, very granular level have a higher risk of being automated, and the ones at the broader level have a lower risk, which makes sense because executing on something small is easier than executing on something large and abstract and complicated. So I think that is just reality of the world for machines and for humans. Doing more abstract, complex things is more complicated, almost by definition. So if we can think of these subtasks as being automated, then the question is, well, if you have three tasks that are a workflow, then why don't we just train a machine or a system to just do that whole workflow? And that's kind of the promise of like agentic AI. So the idea of agenic systems is that they can coordinate.

And we do have that. So OpenAI deep research, chat chapti deep research is a good example of find me everything about some topic. And it goes and it does searches and it does synthesis and it does like it does a set of things to deliver like a finalized report. Claude code is another great example where you can say build a website or which is something. you could do a more complex task. And so AI is moving in that direction of orchestrating related tasks into a workflow. Now, I think there's still higher level tasks

where there's not really any end in sight. So if you could do a more complex workflow, then that becomes the atomic task. And then the human, or whoever sits on top of that, really needs to orchestrate between those sub-tasks. you can manage things that are more abstract and you're like kind of a higher level manager, a director type person,

or you can manage more sub-level tasks. And as machines get better at performing broader tasks, then we as humans, if we're playing more of this orchestration role, we'll be doing more high-value, more abstract work. What I kind of came around to recently is this kind of divide between automation, and augmentation.

Sometimes in economic research, we see some division between, oh, there's some technology that can automate work and others that can augment your job, augment your work, and help you in doing something better. I think these are actually the same thing,

just at a different level of the taxonomy. So I know I'm getting like really wonky here. Let me try to break it down. So at a micro level, if you have a work activity or a work process that has a dozen different tasks, and half of those get automated, then you can concentrate in the six remaining activities.

Your job has been augmented because the tasks have been automated. So I think... So let's say, let's just bring it to an example. You run a cupcake bakery. In the course of a normal workday, there are 12 specific tasks that you do related to, some of which are related to handling the cash register, to managing inventory, flower sugar eggs.

If the floors are slippery because it's been snowing, there is a protocol for some employee to make sure that everything is adequately mopped up so that the customers don't slip. Like you've got all of these disparate tasks that you need to do in the management of your cupcake bakery. And so you're saying if half of those get automated, then the other half that are, are not automated are augmented? I would say like you as a person have been augmented. You as a person, you as a job.

Yeah. Worker. You can do more with your time. So your productivity is enhanced. And that enhancement of your productivity happens through automation. So I'm coming around to the idea that the term augmentation is actually kind of confusing and worthless. It's actually automation of parts of what we do that make us more productive.

Arguably, would that also be the case historically? There are tasks that we used to do that have now been automated, washing clothes. People used to wash clothes by hand. That task has now been automated by a washing machine. Washing dishes, same thing. That task has been automated by the dishwasher. Could you not make the case that what we are experiencing right now is a continuation of this historic trend in which, you know, we used to need to get ice cubes to keep our food.

cool and now that has been automated by refrigeration. Yeah, yeah, it's exactly the same. There's always some part of our work process that gets automated. It's really up to us as workers and humans to reconfigure what we do. We need to adapt. And that adaptation really is a reconfiguration of your work activities. You have some composition of things that you do and it takes up some portion of your time. if things take less of your time,

you need to kind of go back to the drawing board. Sometimes you just concentrate in other things, spend more time on other things that you were doing, things that you like doing, things that are more high value. Maybe you introduce new things. All of that is in response

to parts of our workflows being automated. I think the fear that many people have is that we'll run out of things to do. Yeah. That so many things will be automated because when we talk about refrigeration or the dishwasher or the washing machine.

These are discrete machines that problem solve for extremely discrete tasks. So the invention of the washing machine solved for one extremely discrete task but had no ripple effect on any other task that we did. And I think the differentiating factor with AI

is that AI can automate in one fell swoop or in one new technology. Yeah. Almost everything, ranging from writing poetry, to processing our taxes, to researching how to train our dog. Yeah, yeah, it's extraordinary.

In some way, this is kind of the divide between like narrow AI and general AI. And it's very clear that generative AI, what we think of as chat GPT, is a general purpose technology and that it affects a lot of different things, kind of like electricity is a general purpose technology. It's not doing one particular thing. It powers lots and lots of different things. It won't affect, it's not like an ATM. that will only affect bank tellers.

It is something that will affect every job. And we can talk about the ATM example. That's like a fascinating example. But yeah, it'll affect lots of different jobs, but not entire jobs wholesale. Still, even though it is general, there are different parts of what we do that at best can't be automated.

There's still uniquely human parts of our jobs. But even if we don't want to be that optimistic, at least won't be automated at the same time. Let's imagine that we have this AI that can do anything and everything in the world. As long as those get rolled out at different rates, then there's time for our jobs to kind of adapt. And that allows us to think about what our comparative advantages are to the extent we have them, which I think we do.

Well, I think fear is that in a relatively short, period of time. And this is assuming, because right now the big comparative advantage that humans have is dexterity, physical dexterity, but assuming that we were to get to a point where robotic dexterity could actually keep up, robotic or artificial forms, could keep up with the human body, which we are very far away from right now. Yeah. But 20 years from now, that might not be the case anymore. Yeah. I think the concern is once that happens, then will AI have exceeded us, both in terms of intelligence and in terms of dexterity? What do we have left to do? Yeah. I think dexterity is a very good example of something that AI and robotics cannot do.

Like, you can't have a machine that buses tables, at least not yet. That is still the domain of humans. I don't think it's uniquely dexterity. So there's this wonderful paper. by Isabella Luasu and Roberto Rigabon called Epoch of Jobs or something like that. They have this acronym. Epoch stands for empathy, presence, opinion, creativity, and hope. And it's these five things that are uniquely human. So the idea is that these are in some way not automatable.

Right. Now, I'm sure we can quibble over these things. Maybe machines can get creative, et cetera. but it is still kind of science fiction that we are going to have AI rabbis or AI psychologists, maybe podcasters, I don't really know, but teachers, I mean, teachers are a different story because you can sort of segment like I have two girls in preschool. Preschool teachers, I'm very sure that's not vulnerable to AI, at least not now. It takes so many of these uniquely human things, caretaker, things like that.

Well, so Yuval Noah Harari actually has stated the opposite when it comes to AI rabbis or priest. Yeah. His hypothesis is that any text-based religion, because it is text-based, AI can consume all of the religious text and then can answer any question that you have in order to offer that guidance. Yeah. So his position is that for non-text-based religions, that will be different. But Christianity, Judaism, Hinduism, these are all text-based. Yeah, I saw that he said that. And I disagree because I think it's a misread on what we want from leaders, community leaders.

I don't think the role of a rabbi or a priest is really to weigh in on religious texts in the way that it might have been in the 1800s. it is much more about building community. I mean, they're community organizers and, you know, want to inspire hope and be there for people and create bonds between families. Like, I think it's just a misread

on what the occupation really is. Yeah, it's community leader. So I think it was kind of an interesting example. I mean, I agree that, like, reading text and deciphering text is not uniquely human anymore. But I don't think that's actually what we want in our leaders.

All right. Well, with regard to AI psychologists then, we see programs like Wobot that are designed to provide AI therapy. And I know myself personally, like, there are times that I've been frustrated and I've had a conversation with chat GPT and I've left that conversation feeling much better. I've kind of worked through my frustrations and understood the source and understood how to deal with the situation. And it's been a really nice, it has really served an empathetic role in my life. Yeah. Yeah, no, I think you're right. And also, so many people are confiding in AI for better or worse. There are examples where that hasn't been good. But yeah, it seems like this aversion to AI being someone that you're friendly with is less the case now than it used to be. So maybe there is room for that to kind of take over. There's two problems, as I see it, for AI psychologists. One is that AI, at least right now, doesn't have good memory. So, you know, that's a problem that's being worked on. And there's improvements in increasing the context that AI can incorporate. But psychologists usually don't just meet with someone once. They meet with someone over months or years or decades and really get to know a person. And that presumably affects how they interact.

They can tailor their therapy to what they kind of understand about the full dimensionality of a person. So maybe that will improve. But another is just human taste. I mean, my wife is a child psychologist. She sometimes tells me that being a psychologist for children is so different than like working with their parents in a way that just getting these kids to trust you and confide in you is an art. And that is something that takes social skills. We're not quite there yet with technology to actually get people to open up.

I mean, some will and some won't, but there's still a set of people that feel uncomfortable opening up to AI. How we led into this is the root question of as AI develops, will there be anything left for us to do? And that traces back to... Really traces back to the question of what is a job,

the execution versus orchestration, and then what is our role within that? So I want to take this to a practical level because for somebody who's listening to this conversation, who is between the ages of 25 to 30, relatively at the beginning of their career, trying to understand how to chart their way in this world in which everything is changing so rapidly,

many entry-level jobs are disappearing, and they themselves don't have, at that age, they don't have the best, benefit of accumulated assets that they can fall back on. They don't have experience. They don't have a network. What should they know at this stage, given how rapidly the future of jobs are changing? Yeah. Entry-level workers, young people in general, are in a much tougher spot now than they have been in a long time. One is that it's a very tough environment for hiring. So we're in this kind of low-hire, low-fire paradigm. And that hurts younger workers because they're usually

the earliest entrance into a new hire, but it's also particularly bad in that there seems to be good evidence that the entry-level jobs that are hurt the most are ones that are vulnerable to AI. So it seems like the expectation is that some of these entry-level jobs may be automated or could be automated in the future.

It's unclear whether firms are actually adopted, technology and have less demand for younger workers because of that, or if they're just anticipating that they might have less demand for younger workers because of what they could do with this technology in the future. I think it's probably more anticipatory. But either way, younger workers are in a really tough spot. One issue is this execution and orchestration question where if AI is better at execution, then the return on coordination is bigger. Who has more skills in coordination, it's people that have managed things that have orchestrated before. Younger workers don't really get trained in orchestration. So much of our education is based on skills and

executing pretty narrow things. So not in the managerial. Yeah, right. Yeah, exactly. I mean, even MBA, so I work at New York University. I teach the future of work there. Even at a business school, the classes that are offered are very practical. They're more vocational. And I think if you had gone to business school in the 60s and 70s, that would be the next step to running a division somewhere. Now it's not. Now people finish an MBA and they're a level two analyst at a consulting firm or something.

Management education, I think, has receded, which is unfortunate for this moment in time. So I think there probably does deserve to be a resurgence of management as a discipline because I think it can be taught. What I try to tell younger people is try to work on a project from beginning to end. Take on something complicated and just try to work through the nitty-gritty connections that you need to make from one process and another process. It's really hard. It's surprisingly hard. The return, the relative benefit of being a good orchestrator is hurting younger workers relative to older workers.

So that's, I think, the real issue. The other is just that we are in a risk-off environment, and younger workers generally are a higher variance bet. If employers have kind of a high discount rate and they want to prioritize the present relative to the future, it's a safer bet to find an experienced worker. There'll be more productive day one. Younger workers are just, they pay off over a longer period of time. Yeah, actually, that's also true. I'm thinking about overseas outsourcing. Those are often higher variance hires. Yeah. You know, I was having a conversation about this with a colleague about the tradeoffs between overseas outsourcing versus paying more of a premium to hire somebody who's local. And really, the decision came back to that variance, right, to predictability of result. Yeah, another piece of evidence that we're seeing that is that while we've seen declines in demand for younger workers, we haven't actually seen a reduction in wages for young relative wages for younger workers, which is interesting because you would think, oh, decline. So we see decline in quantity demand. But you think, oh, declines in demand mean like lower prices and lower quantities. But we're actually just seeing those lower quantities at actually even higher prices, which is, um,

on its face a little strange and you think, oh, maybe this isn't a demand thing, maybe this is a wide-side thing. But I think it's consistent with the idea that the safer bets are the ones who are being prioritized. And those safer bets are more expensive. So maybe someone will take a bet on a younger worker who's got great experience, a good pedigree, but that higher variance less certain worker that someone might have ordinarily taken a bet on, they may not be willing to take a bet on in this environment.

What's going through my mind right now is the notion coming from this that if you are the owner of a small business, it actually could, depending on what industry you're in, but if you're in a type of business that needs knowledge workers, it could actually be rational to prioritize paying a premium for more highly trained, highly experienced, sure bet premium workers. It's sort of the opposite of what you often see small businesses do, where small businesses they run on shoe string budgets. They're often bootstrapped. Don't you have a stat, what, 10% of small businesses in the U.S., which equals about 800,000 businesses are solo proprietorship? Yes, yeah, that's right. So if you're transitioning from a sole proprietorship to something with one full-time employee outside of yourself. They better be good.

Yeah. And oftentimes because you're on such a shoestring budget and you're bootstrapping the whole thing, the tendency is to want to hire somebody who's cheap and inexperienced, but it would actually be safer to do the opposite. Yeah, yeah, I think that's right. What we did at Rebellia Labs is we used to hire interns. I think hiring interns,

you get a better return on investment because they're more high variance. It's a long evaluation period, but it doesn't get you that benefit. You basically have to train someone and not really get any output, maybe small levels of output for the hope that they'll join in a year or whenever. It's a very tough bet to make when you're optimizing for the short term.

So we stopped hiring interns. And part of me feels a little bad about that because I know you can get high returns, but sometimes in a startup you have no choice but to optimize for the short term at the expense of the long term. and that is just a choice that every business operator needs to make sometimes. You need to weigh those costs and benefits. And that actually then, so for the people who are listening who are wondering, how do I take this information and process it into the way in which I position myself as a job candidate, having more experience, more managerial experience, and then not underselling yourself,

there is a signaling that if you command a premium rate, you provide premium service. Yeah. Yeah, I think that's right. I mean, there are ways to send those signals, like now more than ever, there's spadging, there's all sorts of ways to do that. But I also think you can send those signals directly through networking and just telling people what you're doing. I mean, kind of evergreen advice for young people is to network more and just get out there, get

people who know what you're doing, and maybe some will like it and maybe some won't. but at least getting that connectivity will, one, show people what you can do, but also give you some sense of what you should be learning and what your next signal should be. You've just listened to Part 1 of 2 in our interview with Professor Ben Zwegg. He teaches a class on the Future of Work at NYU Stern School of Business, and he is the CEO of Reveloabs, a company that uses AI to analyze the labor market. we'll be sharing part two of this interview in our next episode. Part two is longer, and it's going to talk through the taxonomy of jobs. And if you're thinking, why do I care about that? It's because

when you go to search for a job, you need to know what you're looking for. And right now, all of those categories are shifting. So make sure that you're following this podcast and your favorite podcast player so that you can listen to part two, which we will air as our next episode. In the meantime, what are three key takeaways that we got from this conversation? Key takeaway number one. Your job is a bundle of tasks. If AI can unbundle your job, it can replace pieces of you, but not all of you. So the risk isn't that AI replaces your entire job overnight, despite what happened at Block, despite some of the more high-profile layoffs that we hear about in the headlines. For most people, the risk isn't waking up to a layoff. In fact, overall, we're in a

low-hire, low-fire environment. The risk is that AI quietly begins to automate some of the pieces of what you were hired to do, particularly if your role is mostly the execution of tasks, then that's the part or parts of your job they're most exposed, most vulnerable. I think the almost perfect definition of a job is a bundle of tasks. It's not quite the right definition, but I think it's a very useful step forward from thinking of jobs as the atomic unit of analysis.

Sometimes people talk about automation as affecting jobs, but jobs are combinations of things. So I think it's automatically when we talk about automation, we really can't be talking about jobs because we can go more granular than jobs and we can think of jobs

as a collection of things. Your job as a podcaster or podcaster plus, I guess, a collection of different things. You're not just doing the primary activity. You're also preparing and analyzing social media. Like there's all a collection of things. And any person who has a job, they have a dozen or so responsibilities that they do in their job. The job title is really kind of a shorthand for a collection of things. But the elements of your job that are not pure execution of tasks, the more relational elements of your job,

That's the part that's least exposed. So if you think of a job as a bundle of tasks, the most relational elements of that bundle are the elements that are harder to replace. That's the first key takeaway number two. Quote unquote augmentation is really just automation. You'll hear companies say that AI is going to augment workers. Professor Zwegg argues that that often is just automation that's how. happening at a smaller scale.

What I kind of came around to recently is this kind of divide between automation and augmentation. Sometimes in economic research, we see some division between, oh, there's some technology that can automate work and others that can augment your job, augment your work, and help you in doing something better. I think these are actually the same thing, just at a different level of the taxonomy. So I know I'm getting like really wonky here. Let me try to break it down.

So at a micro level, if you have a work activity or a work process that has a dozen different tasks and half of those get automated, then you can concentrate in the six remaining activities. Your job has been augmented because the tasks have been automated. Here's what that means in practical terms. If part of your job gets automated, then your productivity goes up and that's great. rate, but it also means that there's going to be some skill reshuffling because the things that make you valuable start to shift upward. Think about when washing machines came out, washing machines automated laundry. And so this task that used to take hours manually using a washboard suddenly

took minutes. And that meant that the skills that made a domestic worker valuable or a homemaker valuable, those skills then shifted upward. It was no longer about banging a shirt against a washboard. It was about coordination, judgment, prioritization. These are attributes that move up the value chain. And so when we talk about augmentation versus automation, what we're really talking about is becoming more valuable because of that automation. That is the second key takeaway. Finally, key takeaway number three, early career workers are the most exposed to risk, because if you are early in your career, you are often hired in order to execute very narrow tasks. Untested workers typically have a very narrow scope of work, and those roles are the easiest to automate. At the same time, companies are hiring less, and they are prioritizing employees.

that have a lower degree of variance because variance is risk and when as a company we are facing risks externally you can't also introduce new risks internally so the workforce that you do hire you want them to be more established more experienced more proven and that makes this particular combination of what's happening right now particularly tough for entry level workers entry-level workers, young people in general, are in a much tougher spot now than they have been in a long time. One is that it's a very tough environment for hiring.

So we're in this kind of low-hire, low-fire paradigm, and that hurts younger workers because they're usually the earliest entrance into a new hire. But it's also particularly bad in that there seems to be good evidence that the entry-level jobs that are hurt the most are ones that are vulnerable to AI.

The other is just that we are in a risk-off environment, and younger workers generally are a higher variance bet. Those are three key takeaways from this conversation with Dr. Ben Swag. And again, this episode is part one of a two-part series. So make sure that you tune into our next episode in order to hear what next? What do we do about it? How does this affect our next steps? And how does it affect the taxonomy of work? And I know taxonomy of work is not something that you wake up in the middle of the night thinking about. But it's a foundational piece of understanding where you fit into this broader landscape,

especially now that so much is changing. So again, make sure you tune into part two, which is our next episode. Thank you so much for being part of the Afford Anything community. Please share this with the people in your life. I actually, as I was in the middle of recording these key takeaways, I got a text message from my best friend who's worried. We're all worried. I wrote her back right away and I was like, oh, the Jack Dorsey effect. And she was like, yeah.

So if you share that same feeling, first, I thoroughly understand why. And there are millions of people who also feel the same way. And that feeling is grounded in some very real shifts that are happening. And if you do feel that way, first of all, please don't do anything rash with your finances. Don't liquidate all of your assets. Don't thoroughly change up what you're investing in. Don't dive into crypto. Don't do anything rash.

Remember, this is why we have investor policy statements, written statements of here's what we're going to do. So that in times of fear, anxiety, crisis, we can look to the guidelines that we have written out for ourselves. So if you are worried, first and foremost, please don't do anything rash with your finances. Second, remember that acquiring assets puts you in a position of power. And assets means index funds, real estate. Remember, income from rental properties is not going to be affected by AI. While it's not going to be adversely affected, it will make your job as a landlord easier by simplifying the bookkeeping, simplifying the operations.

but at the end of the day, people need a place to live. There will always be rental demand. So buy assets, hold rental properties, hold index funds, because we know that the trajectory of the U.S. stock market over the long term historically has always gone up and then focus on learning about the future of work. Because you may or may not need to retrain. And if you do, and it's certainly. And all of us, whether we have to formally retrain or not, all of us have to pivot and shift and learn new skills throughout our careers.

So be nimble, be agile. And, you know, in the words of Wayne Gretzky, skate to where the puck is going. Pay attention to where things are leading so that you can get ahead of it. Thank you so much for being part of the Afford Anything community. If you want to discuss today's episode, you can do so for free at affordanithing.com slash community. It's where people in this community, afforders hang out. Afford anything.com slash community totally free. We have a newsletter, afford anything.com slash newsletter, where we share insights that we do not publish anywhere else.

So that's afford anything.com slash newsletter. And again, make sure you're following us in your favorite podcast. Player, open up that app. Make sure you hit the follow button. And while you're there, please leave us up to a five-star review. Write a few words. Talk about what you enjoy about the show. Thank you so much for being an afforder. I'm Paula Pant. This is the Afford Anything podcast,

and I'll meet you in part two of this interview with Dr. Ben Zwegg about the future of work.