Why AI Is Changing Jobs Before It Changes Employment with Sebastian Fixson

Speaker A: 2. At that time, PhD students at MIT ran an experiment like that with a task that you could compare to writing marketing copy. So a bound 20 minute type task that involves some creativity, creating an artifact that used to be done only by human and in that experimental fashion. What they show is that if you add AI, in that case as a tool to the person's toolbox, if you will, they show on the task level productivity differences of about 30%.

Speaker B: Sebastian, welcome to the Future Work podcast. We're uh, really excited to have you here today and uh, really to explore the topic. Everybody's exploring, but with your expertise, uh, that's AI and the impact on the future of work. So, uh, thank you very much for joining us.

Speaker A: Thank you for having me. I'm excited to be here, Frank.

Speaker B: Okay, well that's a good stage right there because most people are going, I'm just happy to be here. And you're excited. Oh, that's, that's, that's good. You know, most AI conversations today that we see, the, the big headline ones are about job loss. They're about job loss. And your research focuses on something much more detailed and it's really how AI is changing the structure and the processes inside of jobs. And that's a really important and distinctive point that's different than many people are making. What do you think that reveals on a broad basis?

Speaker A: I mean, it's not surprising given how important work for everybody is, that, ah, there's a big interest in the subject, if you will. But depending on what level you are at looking at this question, you might have a very different zoom, if you will. Policymakers look at economies, entire, um, job categories, industries, company leaders look at firms and their sector, maybe, but individual employees look at their job and these different layers lead often to very different outcomes in the analysis. If you look at the range of predictions of job losses caused by AI, they're enormous. That doesn't necessarily mean that one is right and the other one is wrong. That typically indicates that different factors flow into the analysis. And personally, sort of where my research is a lot more on the ground trying to understand what happens in work. My background is studying innovation work. Um, so my perspective is that's where I can contribute something to the conversation.

Speaker B: Well, you know, as we look back at history and we've all seen a lot of, um, industrial revolution, technology revolution, etc. We've gone through a variety of major changes in the past, which at first were scary to people, but in the end, for every one job, two new ones were created or the value of two new ones were created in many cases. Uh, uh, a single truck in the early 1900s was able to deliver three, four, five times as much material speed wise as a horse and wagon. Um, uh, computers. We know how computers have evolved and how computing power has evolved and how that has impacted us. Same with communications, uh, the invention of the telephone. All these things are breakthrough and they're scary. But they all ultimately have contributed to growth and to improvement. Do you think we'll see the same thing with AI or what do you think the magnitude of the improvement? I guess because you can't not see it with AI. What do you think the magnitude will be? Uh, and I know you've done a ton of research at Babson, uh, with the work Futures Lab. Um, is there material, uh, empirical data that's starting to show this or is the jury still out a little bit?

Speaker A: I think overall the long term jury of course is still out. No one knows. And in fact that's what researchers are trying to do to create projection based on certain assumptions. So if you look at studies that talk about millions of jobs created versus millions of jobs lost, it's not impossible that they'll shape out in some sort of that way. However, there are lots of um, uncertainties around these estimates. That's one even in the aggregate level. But on top of that, this changes, you mentioned from past technological revolutions had very different impact on different people. So if you take the job owner's perspective, the one who loses the job is not necessarily the same who gains the new job.

Speaker B: Right.

Speaker A: And so therefore that humans find that scary I think is completely understandable. That's then where conversations at whether we need job training programs and how different in terms of required skill set and competencies are these new jobs. Um, if you ask for empirical evidence, I think there are two types of evidence that are emerging that point towards large scale changes. The first one are studies that look at essentially in experimental fashions, trying to understand what the productivity impact of adding AI to someone's work is. So a couple of years ago, two at that time, PhD students at MIT ran an experiment like that with a task that you could compare to writing marketing copy. So a bound 20 minute type task that involves some creativity, creating an artifact that used to be done only by humans. And in that experimental fashion, what they show is that if you add AI, in that case as a tool to the person's toolbox, if you will, they show on the task level productivity differences of about 30%, which on its face are enormous. If you Think about it at the same time, in, in the wild, so to speak, the, the so far observed productivity changes are much, much smaller. And that suggests that in reality.

Speaker B: Can I ask a question on that? If they're smaller, is that because

Speaker A: it's

Speaker B: smaller per individual, or is it smaller because the space of so many people that aren't really embedded in the process dilutes it?

Speaker A: Well, that's a very good question. And the answer is it's both. The individual has that task and many other tasks. Right. The job. And as it turns out, in that natural setting, things are a lot more complicated and the contribution of AI, at least initially, is less clear and less clean. Um, the other point you're pointing out, we are in a transition phase. And all of these, uh, adoption and diffusion studies show it's very uneven across sectors, across age groups, across education level. So not even though they are all growing in AI use. And the measure is typically, if I ask you, have you, how frequently in this week have you used AI? And maybe there's category daily, weekly, monthly, but that still doesn't say, okay, Frank, what task have you given to AI, uh, or in which way have you actually interacted in creating the outcome? And that ultimately is the answer to the question what the productivity improvement in real life jobs will be. So right now our measures are so coarse that we're talking about users almost in a binary fashion, yes or no. So we're not there yet to understand what is actually happening to the way in which the work is changed.

Speaker B: Well, you know, it's interesting, um, we have another company, Alliance Virtual Offices, and there's an employee base, about 125 people there. Uh, and the company's been around for a while and the uh, very stable, um, base of executives and managers and staff, very stable company. Um, and what's interesting that I find is the people with the most experience in business, the most gray hair maybe, uh, as well, um, uh, um, are the ones that are reaching out to AI, um on their own the fastest. They're the ones that are saying, aha, ah, I can now do this extra thing that I used to have to rely upon someone else to do, or this task used to tell me, take me two hours, I can now do it in one hour. We're seeing more, um, natural take up, uh, from the people with the greatest experience. I wonder if that's just because they know how long, they know how, how to apply themselves to a work product where a younger generation is still trying to figure out their job a little bit. And doesn't really and doesn't have the confidence to reach out to a new tool as quickly. Are you seeing anything like that or in most cases are we just uh, an anomaly?

Speaker A: No, I think, I think number one, um, as I mentioned, the uptake or diffusion of our tools, um, varies a lot across um, both education level but also leadership levels. And the higher the leaders, the higher uh, the use is. I think one of the explanations for that is that the experience gained through the professional life puts you in a position to judge the outcome with a fairly high degree of confidence. You know, if you get a report from an employee or an output from AI what it's worth, how good is it, how much to trust it? That expertise, however, you have built over decades in your job or in various forms of employment, if you will. And that is actually one of the big challenges for people at the other end of the career arc who are entering a job who would like or uh, who are asked to have that expertise to have good judgment to evaluate the output of AI tools. But they don't have that yet because they haven't spent enough time in this type of work to know how good looks like.

Speaker B: Well, no, I think that in itself is a really critical uh, understanding uh, that people need to have uh, in terms of the judgment capacity. What we've done and what we see is that uh, first we're going full AI native, period. We've decided, we made that decision. Uh, increases in productivity, increases in value as a result of higher employee and security safety for job security and higher compensation through higher productivity. We think we should be giving people raises rather than firing people. Okay, so that's our attitude towards AI so we're going full AI native. Um, and in doing so the first thing we recognized is exactly what you said, is that some job functions, what we need to do is redefine the job itself and then build AI into a process or into a multitude of processes to help the person who may not be able to make the judgments. But we're purely an accelerator as opposed to someone who may be able to make the judgment that's going to be pulling in data and reporting and forming opinions and making higher level decisions. But you really have to reshape each of the job descriptions in order to integrate AI uh within a company, in my opinion.

Speaker A: Yeah, and I would expect that we have to go actually much further because if you think about jobs are functions in, inside of an organizational structure and inside of more or less well defined processes throughout whatever the organization does. And if you to Use the lean terminology, redesign value streams. And if you redesign them in a significant way, you are effectively redesigning your organization.

Speaker B: Well, can you explain what the value stream is in your, your definition?

Speaker A: So in my original home department is operations management. And the lean discussion that emerged 30 years ago in the quality movement and lean manufacturing at its core. If you go back to the early writings, they talk about a value stream. And what that means is that the very first task for a leader is to identify what is the value that we want to create that ultimately customers value, and then align all the activity, the stream towards that goal. And I think, uh, on a very fundamental level, that is still the task that we have today with every organization. However, if AI allows fundamentally restructuring value streams, that means we're looking at very significant organizational change questions. And those of us who have tried this throughout their career have experienced how difficult organizational change is for a whole host of reasons. Right. Legacy cultures, legacy IT systems, people, their own mindset. There's a whole host of questions, but in order to tap into the true potential of AI, that is probably necessary. Um, and if you look at the current discussions that Ethan Mollicks and the kinds of have talking about IT capability overhang that points that out that the narrowly measured the tools are far ahead of what we currently in our current organizational structures can actually realize.

Speaker B: Yeah.

Speaker A: And that change to change the organization, that will still take years.

Speaker B: Well, and this goes a fundamental question I've got, and I don't know the exact number, but I'm going to make a number up because I'm good at that. I'm going to say that 70% of the people in the United States today work for Fortune 1000 companies. Excuse me, 30% of the people work for Fortune 1000 companies. 70% work for small to medium enterprises. And I think that number of bureau labor statistics is pretty close. Um, that means entrepreneurial companies, those smaller companies out outweigh in terms of number of employees. Employees, um, the larger companies. Yet it seems most if not all of the studies are done on the larger companies and smaller organizations. Just like, uh, turning a ship or an airplane or a car, you know, a Ferrari handles a lot faster than a big suv. Okay. Um, smaller organizations can react faster.

Speaker A: Yeah.

Speaker B: Principle too, level of adaptation. So in terms of pure numbers of individuals, I think adaptation from that group will outweigh the others. And it will also happen faster. The other thing I'll bring up is I was uh, recently at a meeting in New York that was a whole variety of investment banking companies and investment funds. And it was really funny because I was speaking in one of the, you know there, there's a uh, there was a little system where people were texting around. He says oh great, we get to listen about the future of work from an old guy with a beard. So anyway it was an, it was an interesting meeting. Um, um but what the one thing that struck me is when you talk about investment criteria for new companies and what the, where the capital, the venture capital is looking. Um, it's not just an AI company but they really don't want to invest in companies that aren't working towards an AI native structure operationally anymore. M they want to see how companies that they're going to invest in um, will be applying AI to gain a competitive advantage. So if this is the case, capital resources will be limited to companies that don't manage this change or limited to companies limited for companies if they don't manage this change. Um, is that something that you've seen or is there any research that you're seeing on that or is it just anecdotal from a dumb ah meeting I was at in New York the other day.

Speaker A: I mean I've on in the investment perspective I've heard similar arguments and the hypothesis behind this is obviously that if AI allows um, much faster scaling then you would assume that the capital requirements to scale are smaller. And so that should be an indicator of the potential of a company um, to grow fast. I think the pressure in an established company is probably a different one to more stock market general assessments. Um, but I think it's also a fundamentally interesting question if the um, financial, the capital raising force that you mentioned is disciplining all small companies in the same way. On a theoretical level you might that think that's true. On the other hand, to go back to, to your um, metaphor of the many small ships, um, if, if, if we compare the AI to, to the wind then yes, every small boat will align with it in the same way. On the other hand, the lower requirement for capital search might also be an advantage for a small company to do things differently than VCs or other private equity people would expect.

Speaker B: Well I think you're right. Um, uh, asset light was the flavor of the month uh several years ago. Still is. It's part of the built in requirement. Now you have to be asset light, have a high level of agility uh, if you're going to seek and gain uh capital um, and new large companies, um, are they tout asset light now? Um, large companies seem to be touting the Benefit they get from immediate AI is a reduction in headcount. Uh, uh, and all of a sudden their quarterly profits look better and so their stock takes a temporary jump. I think that's a band aid, a terrible band aid by the way. It's a reactive structure as opposed to a progressive structure. But we've seen different requirements start being laid out and then they become a standard. Okay. Asset light agility is now a standard period. Okay, now what's the next standard? Uh, AI. Native, uh, AI integration, uh, some level of, uh, new level of uh, revenue to personnel ratios that are different than they used to have to be. Um, how do you see that playing out?

Speaker A: Well, I mean if you look at the newest fable animal, the one person unicorn, there you have your ratio.

Speaker B: That's you, that's me.

Speaker A: Of an enormous, um, ratio of revenue per headcount. Right. Um, but I think directionally that's certainly not. Many entrepreneurs are aiming to create a organization that can do a lot more with fewer people and that. One of the reasons why I'm looking into this problem specifically or this question is, um, startups obviously don't have a legacy. They don't have a legacy in the organization. They don't have a legacy in existing IT systems or accounting systems or um, established power structures. If you form a company from scratch, um, things could evolve in different ways. And at least initially startups um, behave that way. Right. The job descriptions are not clearly delineated as they are in more established organizations. I think the interesting question to me is will certain legal requirements force them to. So I can tell you, I had a, I ran a couple of um, roundtable discussions two months ago and interestingly enough, uh, they're on the table where company leaders and they were in one of two camps. One was that we need to get our people to use AI more and faster. But the other camp said when we are worried about losing control, it's sort of bring your own tools and people do a whole host of things. Where is the data? Uh, what's the risk? Control. Right. Who is responsible for compliance? Those things are unanswered and some of the people at the table were really worried about that problem. And you can see that startups without any experience in those spaces might create organizations where they have to retroactively figuring out to keep control on some of these Items, whether it's IP, whether it's liability questions, etc.

Speaker B: Well, you know the old story about a leader is uh, a leader always has to look back over their shoulder, make sure somebody's still there, I think in uh, um, corporate, uh, development, entrepreneurial development. A lot of times the brain of the entrepreneur is always out in front of the practicality of operations. You know, a lot of times this is true. Um, very few people could go forward and stay operationally attached at the same time. And smaller companies moving faster, this becomes more of a problem. But you know, we've solved that in accounting. We've solved that with technology. We solved that with personal use of uh, devices and all of that. We've solved all of those problems in the past. Uh, is there real fear that we won't be able to solve it this time? Um, or, uh, uh, is this just another time that we just something we have to do, have to deal with that it's an ordinary course of business issue?

Speaker A: Um, interesting question. One of the participants made that comparison and said, well, that's when spreadsheet entered our world and suddenly data was copied from here to there and we no longer knew where the spreadsheet was. And we got better control of that over decades of work and tightening up IT systems. And maybe that is the answer, I don't know. Um, but it's also people in my generation think about information systems as ordered and structured in some way. Who has access to which kind of data, et cetera. But one of the, one of the entrepreneurs that I interviewed, um, has essentially the AI be the memory brain of his organization. They have four people and all user data interviews. They sort of feed into this and each one has conversations with the system for the questions that they're currently working on. And on one level that sounds very organic and very, um, productive. Low barriers of entry. Finding things is left to the device of the AI system to feed it back to you. But at the same time you don't really have, you don't really know where the data is and who has access to which data once. You have to stratify that perhaps, um, in the future AI systems will build in these kinds of control layers. But it's not clear to me that at this point they exist. And some other firms for that reason are very, very careful in handing data, customer data into these systems for that reason.

Speaker B: Well, you know, it's, it's interesting, um, as I look back, uh, um, not too far, um, companies have their own knowledge banks. They have, have, have their own knowledge, uh, that were internal to them, where people could go and find out things, et cetera. Uh, now, um, companies are building, we're building, uh, our own LLM tied to our own company. All of Our own company information, everything in our company is fed into a single bank. And then that our own private model for AI has the ability to access. Uh, now we do segregate that. We do segregate. You know, you can't get credit card information, you can't get certain things that would always be secure anyway. But in terms of everybody knowing everything they need from within the company's history, customer base, customer experience, uh, business plans, et cetera, we think that that's important, uh, overall, because decision making is the most valued element within any organization. How fast can decisions be made and how accurately can they be made and how can they be then implemented? Um, that's what drives companies forward, um, more than anything else. And AI is a massive contributor to that. If it's reading the right information, which is part of your issue, if, uh, it's reading everything, well, it may not be apropos, but if you could keep that funnel on what it's reading, then you, you really can accelerate decision making. I think that's what a lot of companies are starting to do.

Speaker A: M. Yeah. And I think the, the fact that we are in a transition phase, you can see in some other measures. Right. So there are certain model performance measures that develop exponentially. Right. So I don't know if, you know, the company mirror or measures sort of how long it takes an AI system to do a task that a human would take, let's say, 30 minutes, and how much that number is expanding. That is, um, developing exponentially. But then if you, but behind, if you read between the lines, they typically have measures that the system is 50% accurate. They have another one that is 80% accurate, and it's still growing exponentially, but slower. But the closer you get to 90, 95, 98%, and that depends on, uh, reliability requirements of your process, the more the progress looks a lot more linear than exponential.

Speaker B: If that's the case. If that's the case, then I would say what data do you have that says for that same accuracy and time thing, uh, where the human is doing it? Because in my experience, humans are rarely more than 50% accurate themselves.

Speaker A: Fair enough, Fair enough. If that's a comparison, then I think the replacement. Well, the augmentation of humans with AI, um, is pretty obvious. I do think there are processes where you probably need higher reliability. Also consistency over time, you get the same answer. And so those are requirements that in any probabilistic system are harder to meet. That's why the performance progress is slower than in these edge measures, if you will.

Speaker B: Yeah, no, that could make sense. You know, in your imagining work initiative, uh, overall, uh, you start with the premise that most jobs today didn't exist 50 years ago or 100 years ago. Those are all gone. Uh, certainly the way we do them. Um, how does that perspective, that historical lens change the way we should be thinking about AI right now? You just said you corrected yourself and went from replacing to augmenting. Uh, uh, what's your thought on the new jobs that will be created? Um, that we have no idea yet. And what that that is going to lead out for the future?

Speaker A: Well, I think that the fact that we have so little idea of what that might be makes it obviously hard to project or extrapolate them. And if you look in historical terms, um, there have been massive changes technology introduced and used. Right. So for example, Post World War II, when the shipping container was invented, it changed completely how harbors operate.

Speaker B: Sure.

Speaker A: Develop entire, entire logistics chains which the majority of people were physical labor unloading and loading of ships. And now they control cranes, computers, trucks, all kinds of devices, um, to move many, many more goods. Right.

Speaker B: That was the creation of just in time inventory and the whole intermodal system. That was the foundation for the whole intermodal system system. Think how Amazon would exist today if the intermodal system didn't exist. It wouldn't.

Speaker A: Yeah, it probably wouldn't.

Speaker B: Yeah, it could not.

Speaker A: So I think that that tells me two things. One, one is that we have to train our muscle of imagination. What could be right and traditionally in for exact business education, um, in the entrepreneurial realm we do this a little bit, but I think we can do this a lot more, um, helping people envision futures that don't exist and that are further different in a way that from the traditional extrapolation. Like if we add, I don't know, another lane of highways or make the cars faster. Here's what we could expect. And so how do we learn to think? Well, what if something was completely different and there are tools and techniques that I'm sort of exploring and want to test a bit more. What how we can train our students and leaders in being more um, imaginative. But you're right, at that point it's still an imagination. I think the second step to that has to be once you have articulated or identified a desirable future or at least aspects of a desirable future, the following question then is what needs to happen to get there? Or uh, from a different perspective, if you backcast what needs to be true for this future to have a chance. And I think what that means is that it also requires us to think about that the future is not determined, it's being built by people. And so the more we can train people in their own efficacy, if you will, that they have agency in the process, the more confidence they have they can contribute towards a future that is more desirable for them. And in that sense, I think this is a larger question not only on a comfortable level, but on um, a society level.

Speaker B: Do you think government impact on trying to regulate all of this, the way the tools will work, the uh, things of that nature, that um, there'll be a restrictive um, element in some areas, um, that will actually reduce our capacity to take advantage of the technology?

Speaker A: I think that's a good question. I don't have an answer. I'm not a legal expert. But I think that bigger and more important focus is on the outcome. Can we agree on how to um, how do decent, how does decent work look like? What are those elements? Right? And I think we need for instance, better measurements. So, um, just one small example. Um, Sarah Cooper, who writes for the Financial Times in her recent book, has an example of the. What happened to the work of translators, right, with AI and the shift was that the AI did the 80% translation and the agency who divvy out the work. Now I asked the translator, well, can you check? Can you verify, can you polish it a little bit? And as it turns out in that setting, AI has now taken what used to be enjoyable for these translators, right? In other settings, coding is mentioned in another way. The uh, AI took the grunt work and now the problem articulation and the formulation and the checking is actually what the people enjoy more. So a similar replacement of parts of the work had very different outcome on the purpose of the job holder, if you will. And to me that suggests that we don't really have good measures and need a different form of discussing what is the quality of work in addition to the productivity value. Of course, in our economy, something that is utterly unproductive will not survive in an economic sense. It may be a hobby or something like that. But at the same time our evaluation measures are very coarse, right? So if we can add to this some measures of work quality, I think that's where we have to have this sided conversation. I'm not so sure if we can meaningfully specify technical aspects of the AI. Uh, that is a whole another can of worms that I'm not really qualified to talk about.

Speaker B: But it's not a can of worms, my friend, it's a bucket of worms. You know, I. I know, I agree, I agree with you there. Shifting gears a little bit. You know, there's so much talk about flexibility in, in the future of work. Um, flexible work, structures, remote work, hybrid work, asynchronous work, et cetera. All the things that have to do with agility. Um, do you think AI, um, accelerates that or do you think, uh, AI mitigates it? Um, do you have any impact, any ideas about physical work structure as it relates to AI and how that might change?

Speaker A: It's interesting that you say that, um, during COVID I had the pleasure to serve as associate dean for our quality programs. And I was in an administrative role. Um, and that was challenging. Um, I will admit that. And at that time I thought it was very challenging, given the structure in higher education, etc. Now I'm thinking actually Covid was just a training game for us. Yes, I think AI. AI will make bigger changes than we have seen so far.

Speaker B: Yes.

Speaker A: Um, but again, I think the devil is in the detail. So is remote work for every work for every person, for every organization the best mode? I don't know. I think every organization, every person have to figure out what they prefer and what works for, for them. And then the organization has a difficult task to figure out how to put that those people together in a productive way.

Speaker B: I think you're right. Uh, it is very individual or very, uh, discipline oriented for a particular activity. Um, but my view, uh, and maybe this is a bias because I come from the flexible workspace space industry, uh, is that um, some variation of flexibility and let's just call it hybrid, uh, for the moment is here to stay, period. M. Um, and people's. The design of their jobs and the augmentation that uh, AI provides. But the redesign of JOBS will allow a better work life balance as a result of combining AI with hybrid work, uh, etc. Etc. And this is a new way that we will learn to do things with those. The combination of those two together. AI shouldn't just create more productivity. It should also create a, um, better work life balance, uh, as hybrid work creates a better work life balance, particularly in environments where long commitment commutes are part of the process, etc. Uh. So, you know, I think that, that uh, I hope that's the way it works out at least, you know, and uh, without a little hope and a little faith, we don't make much progress anyway. For people that want to implement, for people that say, hey, I want to get on this path, I know about it, I understand it, I'm Dabbling with it, uh, seems to be what's the first major step, the first that an entrepreneurial company or even a large company should be taking. What can we leave our listeners with as the first thing to do to really get on this path in a productive way.

Speaker A: So, uh, let me take an example or one data point from a recent research study that was Covid focused, not AI. But I think I can make the point why I believe the same is true for AI. We followed a couple of companies, how they went through Covid and came out of COVID Same industry, same kind of work, very different responses. Not only did the work arrangement differ that they landed up two years out, but more importantly, the way the employees understood and accepted the new setup different. And so we went back into our data and was trying to understand why is that the case. What happened and what was identify is that even though the companies are in the same industry, they have very different underlying cultures and very specifically different ways of involving employees and finding the new, let me call this work arrangement the hybrid work and flexible work. And that made a big, big difference in the way the companies now in the post Covid world operate. So taking that insight into the AI world, I would say do the change with your employees. Not only do they know much better in, um, detail on what work needs to be done, but you also want to have a system on the other side of it that everybody agrees on and is happy with. I'm not saying everybody needs to be 100% happy, but overall use the creativity to create a system that people enjoy working and that generally comes also with B productivity.

Speaker B: Anyway, yeah, no, I, I think that, uh, middle going both ways in a company, uh, is better than top down, bottom up is better than top down. All these ways that are inclusive, um, and gain the, I won't say consensus, but the understanding. So actually, transparency, transparency of benefit, uh, to the customer that you serve, to the people that are on the teams and the employees that are inside of the company, to the shareholders. Transparency of benefit, uh, is critical. It's critical. Well, Sebastian, thank you so much for your time today. I know you're incredibly busy with everything that you're doing over at Babson. You've got amazing work that you're doing from a research point of view and I'm just so grateful to, to you for joining us today. Thank you.

Speaker A: Thank you, Frank. It's my pleasure.

Speaker B: Take care.

Speaker A: Bye bye.

Speaker B: If it's impacting the future of work, it's in the future of work Podcast by Allwork Space.