Human vs Machine: Navigating Careers in the Age of AI | David Mattin and Tad Smith (Round 2)

0:00Hey, everyone. As you know, on this podcast, I bring the best guests in the world of that nexus of understanding of macro crypto in the exponential age of technology. If you're enjoying the show, a quick five-star rating goes a long way. It helps us grow and keep these conversations coming with the best guests in the world. Thanks a lot.

0:49Hi, Exponentialists. Hope you are all well. I am back on the platform. I think it's only been a matter of days. It's been three or four days since I last saw you, so I trust you're well. Now look, as you can see from the title of this one, we are back for round two of a project I just love. It's 2035, an operating manual. And that is a series of conversations with my colleague and good friend, Tad Smith. Tad is the former CEO of Madison Square Garden, former CEO of Sotheby's. He's a partner at 10T Holdings. He's a big voice on crypto Twitter or X. He's all kinds of things to all kinds of people.

1:29And we are going to have a second wonderful conversation in this session. Tad, it's so great that you're joining us again. How are you? David, I'm terrific. And given that we're going to talk about work today, I should probably point out that Tinti Holdings' name has changed to 50T Funds, and everything I say does not reflect their views. So the good hygiene checkbox. That is noted. That is duly noted. And yeah, I knew that, right? I've heard this. You told me that last time, 50T Funds is the new name. Now, look, as you say, we're going to do work in this session. So 2035 and Operating Manual is this journey we've decided to go on into our shared future, into the collision of these incredible technologies.

2:17Of course, AI is perhaps the biggest. Reshaping the future, what it all means for us, how we navigate it as investors, how we navigate it as professionals and human beings. And we're going to walk through a series of topics in this journey. And we've decided that this conversation is going to be all about a topic that is hugely on everyone's mind, which is the future of work in an age of intelligent machines. And there is so much to get into here. Just a little context before we just go, go at it. Everyone is, of course, asking, what does AI, what does intelligent machines mean for the future of work, for the meaning of work, for my career personally?

3:04These are huge questions. So far, any kind of slam dunk effect is clearly absent. If you look at white collar employment as a share of total employment in the United States today, it's stable. It hasn't really changed since the advent of ChatGPT, but there are early signs that that is starting to shift. And I just want to look at three very quickly. If you look at a new benchmark OpenAI launched just last month called GDPVal, This is a benchmark that tests how good our AIs at frontier models now at doing economically useful tasks. So most benchmarks so far have been about fairly abstract intellectual problem solving.

3:59This is how good our AIs at doing economically useful tasks. And they've taken kind of real tasks drawn from real people's work days and asked AIs to do them. And look, to cut the story short, what you see is that frontier models are getting better quickly at doing economically useful knowledge work. They're getting better in a linear fashion, but it's a pretty steep line. They're getting better fast. What's more, if you look at some research done by probably the leading economist on all this, Eric Brynjolfsson, who studies the impacts of technology on employment, on productivity. he's kind of dug a layer deeper and it gets really interesting because what he finds again to cut a long story short his his paper is called canaries in the coal mine and what he finds is that among in domains that are particularly vulnerable to the kind of ai we have at the moment so content creation, customer services, some others, hiring of entry-level positions is diminishing rapidly.

5:09In other words, a strong signal that corporations, organizations are, instead of hiring young grads and entry-level people, they're asking AI to do that work. That's corroborated by some research done by a couple of Harvard postgrads. And I'll put all these links into the chat. And they've really dug into the data and found that companies that are embracing AI, the companies that say they're using AI are hiring fewer entry level positions. So, you know, headline for me is, look, it's not happening yet full pelt, but the early signs are that it's starting to happen and we're just at the beginning of something monumental.

5:58And we from there can just roll into it. And Tad, I have all kinds of issues I want to get into with you. And I think I start with, you have this incredible background as a CEO, as a leader inside large corporations. What do you think corporations look like given all this like five years from now, 10 years from now? And how would you as the CEO of Madison Square Garden of a big corporation be handling this moment if you were the man on the spot kind of dealing with this now? Yeah. Good questions, Paul. So a lot of how, Let me do your first question first. I struggle to see how companies five and 10 years out have the same level of employment as they do today.

6:57And I suspect what we're going to be experiencing for quite a while is the paradox of GDP going up, productivity going up, and overall labor employment going down. And there are lots of ways to think about it. But one thing, one version circulating is that you look at companies with significantly high labor content, particularly now in the knowledge space. And you look at that and you say, your labor content is my margin of opportunity from a competitor. And that is going to be a significant transformation. We'll talk about the social effects of it in just a minute. But that's really, that's where I see five and 10 years out.

7:5310 years out is a little harder to see, but five years out, I can't imagine that it's going to be anything where you have as much labor content as a percentage of the total revenue we should do now in almost any company. The second point I would say is, and you mentioned this, you called it content, I would call it both content and also software development, where you see it particularly depends on the type of company you're operating in. Right now, code is increasingly able to be done by agents, and the AI itself is beginning to do it. So where you'll see significant labor erosion, you're already seeing it in those areas.

8:33And I suspect it will move to more experienced jobs very rapidly. And that also, by the way, is in the link you're sending around. It's a very powerful view on it. So let me just pause there and say everything else being equal, I see companies having fewer people working in them and reasonably soon. Also, we should talk about, obviously, we'll talk about a little bit later in the conversation, what effect that will have on wages and the total amount we spend on it, too. If I were a CEO today, again, it depends on which type of company I'm in, but I would be designating a top executive with AI responsibility in the company to figure out all the applications.

9:20There is a chance if I were a large company, I would be getting expert advice on it. There would be board committees that are either being assigned or potentially reviewing it. And there would be metrics and goals put around it because almost any aspect of a large company's operations now is going to be influenced by AI. And I think the competitive intensity is now visible on the horizon. The one thing that I think I wouldn't be doing or I wouldn't recommend anyone doing if I were CEO is doing nothing. For sure. That feels like the biggest risk right now would be to just try to ignore this. You know, yeah, this Harvard study is really interesting.

10:04And it takes a cut of the companies that are trying to do something with AI and uses those that are doing nothing with it as a control group. And my first thought when I read that is like, who the hell's just sitting this out? Who's doing nothing with it? Well, there are some. There are some. And by the way, remember that this one, unlike the internet, the internet to some extent was a precursor in the late 90s. But this one entails a huge amount of cognitive dissonance. And I think you're seeing that in the labor data. You yourself said that the labor data is putting a lot of pressure on entry-level work.

10:41but it's not necessarily putting much pressure or at least measurable pressure on mid-career and upper-career individuals. And part of that, I would say, is in an era of uncertainty where we don't fully know the impact and the models, although, quote, there, you need extensive experience to make sure that what they can do is ready to be deployed. There's not a lot of upside, to be frank, in corporate leadership, replacing a lot of years and years and years of experience on the promise of AI. It's much, much easier decision to limit intake, limit hiring, and begin to apply it there. And also to move people around, particularly people that have more translatable skills within the company and slow the spigot of hiring much, much easier.

11:39And remember, these are CEOs, by the way, that are fresh from the pandemic where they couldn't hire fast enough and they couldn't find the labor that they wanted. So they have a recency burn, if you will, from that experience. And then bang, they have this superpower technology that It can replace significant tasks of their workers, what they do. That technology is evolving. So it's not perfect. And moreover, insofar as they don't have enough experience with it, they maybe tried it and they saw a hallucination or they read it in the New York Times last year or whatever. Unless they're immersed in it, they don't know how replicable it is to replace tasks of their existing people.

12:20So there will be friction. And there is. For sure. Yeah. And it's really interesting that those dynamics are clearly how this is starting to play out. In other words, it's playing out first with limiting of entry level kind of hiring. And just in my own experience, that makes sense. I mean, the big organizations I have lots of dealings with, you know, some, for example, Accenture. Just, you know, I'm very familiar with the kinds of work that goes on there. And the way AI can help, can most sort of meaningfully, concretely help that type of work right now is you can have mid and senior level staff saying, hey, like, I'm just going to use AI to give me a first draft of this report, of this PowerPoint deck, of this analysis, of this talk I'm going to deliver, whatever it is.

13:08And then I will edit it up into shape, right? So whereas once you had like an army of juniors doing that, now that's kind of, you can just get a really decent first draft. And then you need someone with the taste and experience and knowledge to kind of make that good enough. But that kind of okay first draft is what your junior used to do. Yes. So two interesting things come out of that, right? Like one, well, to me, one is what are the key challenges for that in organizations? The obvious question around that is like, where do you, where does your talent pipeline go? Who becomes your senior people if you kind of kill hiring of junior people?

13:53And the perhaps even more kind of powerful resonant question for most people is, what does this mean for young people starting to navigate, you know, enter into? I feel like it must be terrifying to be graduating into this environment now. We're going to talk about education, so we'll get a lot into this, into what should we tell the kids, basically, in our next conversation. Yeah, and I teach. That's what I'm saying. We'll talk about that, too. We'll get right into that. But how are your young people in your classes experiencing this moment? And what do we think it's going to mean for them for their first decade of work?

14:34So the first thing I would say is that anyone who thinks young knowledge workers are blind to the trend, or at least the noise around the trend, is simply not switched on. they're well aware of what's going on and and um they interestingly you have two different sort of almost a separation if you will between those who are embracing it diving deep into ai learning how to use it becoming incredibly facile with it and viewing it as a life-changing tool and potentially helpful to their careers and those who are doing things the way they were taught to do them without AI and resistant to it. And when I say resistant to it, they're just not in the habit of using it.

15:32And because they're not in the habit of using it, they are increasingly separated from those who understand what the power of it can be. And I'm talking still about the young cohort of folks. And that's deeply concerning, With respect to the notion about what happens, to your other point, about the talent pipeline, well, remember that the premise of your question, the talent pipeline, is that in the future, they'll need a talent pipeline. So a quick break in your regular programming. If you're serious about your future, grab my free report called Prepare for 2030. I think you've got five years to make as much money as possible.

16:19And this guide will help you navigate what's coming. The link is in the description. Download it now. That does underpin my question. And I suppose I still feel intuitively that my kind of overarching take on this is in alignment with you. It's very hard to see how corporations don't lose a lot of junior and middle-ranking, middle-level kind of knowledge worker, like human staff.

17:09clicks. Feel ready? You can move to real money with as little as$100 once your account is approved. And the great thing is that in addition to crypto, Plus500 gives you access to a wide range of instruments. S &P 500, NASDAQ, gas, and much more. Explore equity indices, energy, metals, forex, and beyond. With a simple and intuitive platform, you could trade anytime, anywhere. Experience the fast, accessible futures trading you've been waiting for with Plus500. With over 20 years of experience, Plus500 is your gateway to the markets. Visit us.plus500.com to learn more. Trading in futures involves the risk of loss and is not suitable for everyone.

17:46Not all applicants will qualify. Plus500, it's trading with a plus. And I think certain kinds of corporations, partnerships, might start losing senior ones too because they see the handwriting on the wall and they start wanting to position themselves in new slots where they still can. Right, yeah, yeah. But the organization will still, I suspect for some time to come, want or need some form of like very senior human representation out onto the world. For some time to come, I think that's true. And by the way, there are some aspects of that that could be quite a long time. Yeah. Meaning ambassadorial roles, curation roles, public face roles, sales or problem solution, the human aspect of that roles, all of those roles will be tremendously valuable.

18:40Leadership roles. Exactly. By the way, not to mention AI deployment and strategy roles. Right. Exactly. Exactly. Exactly. And I guess, you know, in terms of that division that you talk about is really interesting between the young people who are willing to kind of embrace these AI tools and those who are just sort of resistant to that or just less willing to engage. And I think in terms of actionable takeaways for people listening, that feels to me such a clear divide. As we stand on the eve of this entirely new work environment, I think one of the big dividing lines is going to be between people who are able to engage in this stuff and use it and supercharge themselves through it and become more effective operators because of it and people who resist.

19:33And it's going to be very, very hard for people who resist. So in terms of navigating this work environment, like point number one for me is, you know, start using these tools, start finding the ways these tools make you better at whatever it is you're bringing. A hundred percent. Everyone should be trying them, embracing them, by the way, and not uncritically using them, to be sure. Let's be clear on that. But critically using them, that is to say, thinking about the best that they can offer and where they fall short, supplementing them with that human element and a good smart eye and judgment and all those other things.

20:20Um, uh, the last thing you want to do is submit a legal brief that you haven't read that AI cooked up. I mean, and one to use one interesting example from the news this year. Um, those are things that are just no-nos, but, um, uh, everyone would benefit from understanding what they do, understanding their power, becoming more facile with them. It's just, there's just all upside, no downside in it. And I have further questions about sort of what you're, what you're telling the young people in your class about and and and other you know other young people you know kind of the children of friends people like that before i get to that like here's a counter thought and it's one we've discussed before you know this idea that um tasks are not the same as jobs and ai is great at like eating specific tasks okay yes but it but But it's rarely able to eat enough tasks and the exact right combination of tasks to eat an entire job.

21:25And a great example of this, one we've discussed, by now I think you can call it a classic example, is the MRI scan. Like AIs have been great at reading MRI scans for quite a while now. They're better than human beings at reading MRI scans for quite a few years, I think it's fair to say. But we have more MRI medical professionals than ever before. We're still hiring them. And that's because being that job, being that medical professional is, yes, about reading the MRI, but it's also about a range of associated other things like communicating with the other doctors, communicating with the patient, just being a human being in that process in all kinds of ways.

22:06And AI can't do those things. What if that is the way, actually the way, what if we're overblowing this and that's actually the way this is going to play out and we're not going to see a huge impact on employment. We're just going to see everyone using AI. Well, I respectfully disagree with a part of that, which is I think the area where we could be dead wrong is in one of two points, which is the AI model suddenly hit the wall. or number two, the paradox, which we'll talk about later, about as new technologies come, they create jobs that you don't anticipate. With respect to the MRI situation, I don't think that's a good example for us to hang our hat on.

22:51And the reason is, is because MRIs are part of the medical profession, which is very heavily regulated, requires a licensure, and certainly in countries like ours has a lot of legal liability related to it. And in that instance, you need a human in the loop who has a license in order to execute the task. The fact that significant portions of the task are hollowed out, and also the fact that that license is controlled tightly, by the way, both by the states and also by, frankly, the medical profession, you can imagine it will be highly resistant to replacing workers for quite a while. And the closer I would think generally, the closer you get to government, government licensure, government protection, regulated entities, things like that, the more resistant you're going to get to broad transformations in the replacement of labor.

23:49That said, going one level deeper, we should not rest on our laurels with the idea that only 10 % of my job can be done better by AI or 15 % of my job. Because in a capitalist system where there's tremendous pressure on returns and lots and lots of creative destruction using a Schumpeter's term, what you find is that people very quickly reconfigure the entire pool of tasks so that it can be more efficient with fewer people that's the rub so i wouldn't i don't think there's a place to hide there instinctively i feel that's right like short term i think it i think some of the effect i'm talking about will play out medium and certainly like yeah exactly you know, like the bundles of tasks will be rearranged such that humans can be knocked out, you know, and you, I mean, again, take just - And by the way, with the licensed version, let's go back to MRIs for a minute.

24:56Soon, I predict there will be someone who does not use AI and gets the wrong answer when AI could have gotten the right answer on an MRI scan. and then that will set an interesting court precedent and that will set a new notion. And by the way, that could potentially transform things very rapidly. We implicitly assume AI produces a worse result than the human being and therefore we want a human being in the loop. The value of the human being in the loop in that medical example that you gave is that that person has the license and that person's legally responsible for the diagnosis and is also authorized by the state because they have a certain level of qualifications and they've reached a number of hurdles to provide the diagnosis rather than, say, chat GPT's scan.

25:47But don't think for a minute that a clever plaintiff's attorney won't find an opportunity to say, well, wait. Whether that person is designated or not, the person had to use AI because they had a duty to the patient to do so and they didn't. Right, exactly. And that's where I think this is going to play out in all kinds of sort of lumpy and unexpected ways because there'll be some jobs where the bundle of tasks really is 90 % replaceable by AI pretty quickly. And that job just goes. There'll be other jobs where that's much more awkward and difficult. And then there's regulation involved. There's legal liabilities involved.

26:25There's some people's job is basically going to end up being to be the person who's legally responsible for things, for the decisions taken by AI. Yes. And by the way, As you apply AI to a human-centered analog enterprise or even a human-centered digital enterprise, what you're going to find is that they're new tasks. How do I intermediate between the human being and the AI? How do I put AI in the loop? We'll create all sorts of tasks and complexities that will also create opportunities for human beings. Yeah. And that actually pretty naturally takes us on to this second counterthought, which is a kind of Jevon's paradox counterthought.

27:05So people have talked about this in relation to kind of compute, you know, an NVIDIA, this idea that actually making compute more efficient could be great for NVIDIA because when you make it more efficient, paradoxically, you sort of increase the demand for it greatly. and people end up using even more of it that offsets the kind of cost decline and nvidia ends up making even more money basically what if you see something comparable here and ai makes human labor radically more efficient um lowers the costs of services that humans provide so greatly increases the demand for those services because human beings are just insatiable and it feels like we're sort of in late stage capitalism and it feels like all we do is buy things from one another, but perhaps we're just scraping the surface of that and there's so much more we can do for each other and buy for one another.

28:04And so actually, paradoxically, you end up with more demand for human labor than you did before, especially given that managing these AIs and these tools and all the rest of it will create new kinds of jobs that we don't even know about yet. How do we feel about that? That feels to me a more persuasive, longer term pushback to this idea that AI just devastates work for humans or the job market for humans. What do you think about that? Well, I think there are two things I would observe. The first one is, can we imagine that as the technology rolls out, there are periods of time where human beings can more flexibly and quickly take certain things on than machines?

28:59One, we just talked about how you integrate AI into a company to make it more efficient. That's hard for a machine to do. That's a much easier task for a human to do. And that's a new task, as it were. And but I don't think. At the speed, these things are evolving and developing, we should expect those tasks to remain like that for very long, because, again, the premise of the question, in order for that to be sustainable, that human beings will have tasks that they can do better. it's the presumption is that the machines are not evolving very rapidly. And we haven't even gotten to recursive creation, meaning where the machines are improving the machines to do tasks.

29:48So, yes, it's possible that we'll experience a period of what I would call AI friction in which human beings are rapidly redeployed to new areas and things, such as this imaginary job of bring AI into a human enterprise and make it work better, which undoubtedly is already happening. That's a new job. But I am cautious on the idea that it will be sustained very long. The second thing, and this is more sobering, if that weren't sobering, and I don't mean to be too sobering today on a Monday, but we need to think about the effect of that on wages and compensation. Because as the demand for labor grows slack, wages will become lower and lower and lower and more and more depressed because you'll have more human beings that are enthusiastically embracing things to do.

30:50So there'll be a lot of social consequences and political consequences and economic consequences of that, even if you're right. Yeah, yeah. And it feels like we at the conversation now can enter kind of a new a new chapter or this book, you know, starts part two because we've established the big picture. We've laid out some challenges to the idea that AI is going to lead to radically diminished need for human labor. And we've kind of, we're not persuaded by those challenges. So does this amount to saying, you know, I opened part two with the thought, does this amount to saying a huge human problem is approaching that is going to play out very painfully in the lives of, you know, tens of millions of people, the economy, the world of work, the importance of of labour is about to be radically transformed in ways that are very difficult to understand, very difficult to imagine what society looks like after this shift and are very painful for lots and lots of people because they won't have a job and the people that do have a job, wages will be severely depressed as you say.

32:04Is that how you... I know you don't want to be too doom and gloom, especially on a Monday, but it kind of sounds as though we can't avoid that that is where we see this heading. Well, keep in mind that we also see cancer being cured. We see wealth being created at a substantial rate. We see human productivity beginning to soar. We see people that are living, having vastly more opportunity to live at higher standards of living. We see things like distressed governments able to provide for their social programs as wealth is created. We see substantial debt repayment in a lot of the Western countries that are struggling right now to make ends meet.

33:03We see advances in science that are at the moment unthinkable. We see much greater human efficiency on things like energy and the deployment of the grid. We see that whole thing being challenged, fixed. We see better situation with those who are concerned about the climate. We see abundance. And the challenge and the opportunity for us is that we need to manage both at the same time. And so I think we can have a sort of dystopian take on it, or you can have, and I don't want to say utopian take, but I can say we're going to get both. And how we manage them is important. In this particular conversation, I think we can think about those as possible scenarios of the future, and maybe they're both scenarios in the future.