AI is expected to have a profound impact on the employment landscape in the coming decades. Kim Parlee discusses the investment implications with Bill Priest, Executive Chairman and Co-Chief Investment Officer, and Kevin Hebner, Global Investment Strategist at TD Epoch.
Kim Parlee: Our next guests have come out with a new paper to talk about how Artificial Intelligence, AI, is going to change how we work and how that is going to dramatically change the opportunities that lie ahead for investors. So much so that they state that AI is destined to be the key driver of equity markets over the next decade. They say it’s the fourth wave of digital technology after PCs, internet, and mobile and call AI the new macro.
Joining me now, Bill Priest, Executive Chairman and Co-chief Investment Officer at TD Epoch, and Kevin Hebner, Global Investment Strategist also at TD Epoch. Hello, gentlemen. Nice to have you both join us today.
Just to let people know, we’re going to be going back and forth between you. But I’m going to ask Bill to set us up, if we could too, and just give us an overview of the thesis. You talk about, in the thesis, that there are four key implications. Just maybe briefly tell us what they are so that we can dig into each one.
Bill Priest: Well, essentially, I think what we’re going to see is a significant change in the profitability of corporations in the sense that if you can substitute technology for labor or physical assets, you can improve your return on those assets. AI will facilitate that. So the way to think about this, if I can hold my revenues constant and substitute technology and technology enhanced by AI, I will have lower labor content and I will have higher profit margins.
Similarly, if I can also substitute technology and to some extent that could be enhanced by AI — if I can substitute that for assets, my sales per dollar of assets will go up assuming I can hold my revenues. AI accelerates this, what we call, substitution of bits for atoms. And this has been going on for a while.
But the outlook for companies that can embrace technology as a substitute for labor and physical assets, they are going to be in good shape. And one of the questions we like to ask of management, in addition to how do you allocate capital, you need a business strategy for the digital age, and AI makes that even more important today than it did before.
Kim Parlee: There’s a lot of things in that statement, Bill, that are pretty loaded. And I know we talk about the labor market, the efficiencies that we’re going to see, although — we’ll get into the implications in just a second — productivity, sector concentration, and the free cash flow generation that you’re talking about. How far along do you think we are in this AI journey? Is this just the beginning?
Bill Priest: It’s just the beginning. In many respects, I think Chatbox, GTV is where the iPhone was when it came out in 2008. We are in the very early stages of the application and usage of AI.
Kim Parlee: Kevin, bringing you in too, let’s talk a bit about the labor market. Bill mentioned that, and let’s just talk about how AI will be disruptive to the labor market. I think a lot of people are scared there’d just be a lot of jobs will go away.
And I actually think what’s quite surprising when I look at this is you actually talk about the impact — maybe this is medium term, longer term — is you’re going to see overall employment and real wages to rise. But maybe just take us through how you see AI impacting labor.
Kevin Hebner: Thanks, Kim. On your — on your question earlier on where we are in the development of AI, actually one of our traders was asking me that just this morning. And we’re still in the top of the first. So it is just getting going.
In terms of the impact on the labor market, it’s going to be highly disruptive. You can think that for about 80% of workers, 10% of their tasks are going to be directly affected by AI. And this is similar to what’s been seen since 1980, for example, with the IT wave.
And more so for about 20% of workers, you’re going to have 50% of their tasks directly affected. So that’s a lot of workers. In the US, that’s about 30 million people. In Canada, that would be about 4 million people that are going to have more than half of their job directly impacted by AI.
So it’s a very big deal. But it’s very similar to the process that we saw from 1980 with IT, or going back further from 1890 with electricity, going back even further from 1790 with the steam engine. So we’ve been through these waves before, and each time, there is a lot of dislocation.
This might be more difficult because it’s happening faster, but ultimately there will be lots of new jobs. And the sectors that are being most affected, for example, healthcare is one of the most affected sectors. We think healthcare employment over the next decade would actually go up about 20%. It’s already about 13% of the US labor force.
So there’s a lot of room for that to increase, education. And when you think about it, doctors, nurses, teachers, it’s not like we think we have too many of these now. So AI offers this nice tool. It’s expertise in a box to help people become even better they are at their current jobs. And then, jobs will change a lot with the advent of AI.
Kim Parlee: It’s interesting, you have a list within this report. You talk about the jobs and you mentioned some that would be affected. I think one of the things that I really was struck by was you have a quote in the article, I believe from Larry Summers, who talks about a world that’s been defined by IQ, your ability to — cognitive processing is that it’s going to be more important to have EQ than IQ.
I think you put that in there for a discussion, if people kind of see if they agree with it. But tell me about that more. Does that just mean we’re outsourcing our smarts to AI and it’s just going to take care of it all?
Kevin Hebner: Well, I think it’s interesting. So for example, if you think education — and it looks like education is going to be disrupted, the most education has been disrupted maybe since the time of Athens because we’ve been doing chalk and talk for a long time. But essentially with AI and things like Khanmigo that Khan Academy are offering, each student gets their own tutor. Each teacher gets their own teaching assistant. And the way that we teach and the way that students proceed through lessons and the way we structure that can change quite a bit. And that should be terrific.
Healthcare is another sector that’s going to change a lot. So we’ll have radiologists working with AI. And because they can see a lot more images, you’re going to see an increased demand for radiologists because we’re going to learn a lot more for that. Doctors, one of the early applications, many doctors spend two hours a day on transcription, so speaking about what they want to do and then transcribing it. They can have an AI tool do that in 15 minutes. So that means that they will have more time with patients to do their diagnoses and do their explanations with patients. So ultimately, this will be playing in, certainly with EQ, but we do have this expertise in a box with AI and the things that we can do that AI can’t. Certainly, EQ and empathy is at the top of the list.
Kim Parlee: Kevin, I want to talk to you a little bit though about productivity. Obviously, the things you mentioned in labor will lead to more productivity right now. But you’re looking at, and I believe, that you could see an increase in productivity by 20% over the next two decades.
Kevin Hebner: Yes, and for backdrop, productivity is important because ultimately, productivity is what drives wealth and prosperity, and what drives productivity in terms is technology. And particular, for the last 300 years, we’ve seen an increase in wealth, prosperity — prosperity has been driven by technologies like the steam engine, electricity, IT, and now with AI.
Early academic estimates suggest that AI — AI, as it currently exists, will increase productivity by about 20% over the next 15 to 20 years. And they get at that by a number of ways.
One is by looking at historical examples where this has happened. And the number ends up being quite similar to that. In fact, the trajectory for productivity post-electricity and post-IT are almost identical.
But the view is that for about 60% of workers, we’re going to see their productivity increase by 30%, and that gets us to a number close to 20%, which would be roughly one percentage point a year on top of the government’s estimate of productivity growth about 1.5. So that’s very important. And there’s a number of easy lifts.
For example, customer service workers, early experiments showed that their productivity can go up by using a large language model by 35%. Software coders, it can go up by 50%. People who write for a living, their productivity can go up by 50% to 100%. So there’s been a lot of early work on this, even if we are just in the first inning of the – of rolling out AI.
But a lot of work where we’re going to see the productivity improvements in many different sectors. So it’s really a broad effect. It’s diffused across the economy.
Kim Parlee: Just a quick follow, you give an example in the paper, as well. And mind you, I’m taking probably the more histrionic example, if you will. But the coal sector, when they added in, I think it was productivity of 832% to the point where there was no more people doing that work, albeit dangerous.
But that rate of change can happen quite quickly. Is there a concern that — that rate of change could just cause a huge dislocation?
Kevin Hebner: Yeah, I’ll do you one better on the history side. So for example, from 1790, if we’re having this conversation and I told you that technology is going to destroy 90% of jobs, some people would be terrorized. But that is what happened. And most of those jobs were in agriculture.
And one consequence of the different technologies is people got different types of work. In most cases, they were better types of work, better paid, better on their bodies, and good in all sorts of ways. So I think you mentioned the speed of technology, and I think that is a concern because with the steam engine, it took about 80 years. With electricity, it took about 30 years. The internet was a bit faster. Our view is that AI will have 50% adoption within 15 years. So it is faster, and that means the dislocated impact will be more challenging for quite a few people.
And there will have to be efforts by governments and companies to help people with that process and with the skills they need. And if you can think from 1980 when we had computers introduced, a lot of people didn’t know how to use these. But within two to three decades, we had 80% of workers using computers as a core part of their job that required a lot of training.
And we’re going to have to have something similar with AI because in 15 to 20 years, 80% of people, AI is going to be an important part of your job. And there is going to be a training element required for that. It’s also going to be, a lot of people need just to go home, go into ChatGPT and just start playing with different prompts and see what happens and try to understand it. But we have to understand this technology. It’s coming. It’s going to affect the vast majority of us, and overall, it’s going to be a very good thing.
Kim Parlee: It is. I play with ChatGPT all the time. You should see the limericks that I write, songs too. But anyway, Bill, I want to come back to you.
The fourth implication, I know that we were speaking about with this paper, as well as this winner takes most dynamic. AI is incredible, but it’s not the small mom-and-pop place that’s going to be developing it. These are big players that are going to be doing it and they’re the ones who will be reaping most of the benefit.
Bill Priest: Yes, in many respects. Again, it gets back to this fundamental principle of substituting technology for labor and physical assets. To the extent you’re a corporation with scale, the ability to do that is going to seriously enhance your profitability.
And Kevin was alluding to this. When you look at what drives real growth, there’s only two things that really matters. It’s growth in your country’s workforce and productivity. Well, the workforce in the developed world is going to be growing very, very slowly — less than 1%, probably closer to 50 basis points, with the exception of Africa. So you’re going to have to look for productivity in the form of the application of technology. That’s actually going to accelerate, as Kevin alluded to.
So, from a country standpoint in the developed world, you can still potentially look at real GDP growth of somewhere maybe as much as 2% — not as much as it used to be, but it’s going to be there. Now, some people will be affected by this more than others. We have some companies we can talk about in that regard.
Kim Parlee: Yeah, we’ll get to the countries, but it is interesting that you’re right. It’s the companies that have the ability to switch over, to substitute, as you’re talking about, the AI, and for the other inputs in there. Are we going to see — is this going to create fortitude around some of the big players that are launching AI, or are we going to see some titans fall, perhaps, in the next little while?
Bill Priest: It’s probably too early to see some titans fall. I think it’s very important to recognize that we are in a very different age. We call it the digital age. And there are four strategies that are going to be required for success. One of them is what the subject for today is, is what’s the business strategy for generative AI? But there’s also — you need a business strategy for the coming energy transition. You need a business strategy for what we’re experiencing, in terms of deglobalization, or think reshoring.
And again, you’ve got the aging of society. How does this impact your ability to allocate capital inside a business? But right now, the subject of today is AI. And it’s powerful, it will be a very powerful influence.
Kim Parlee: You’ve been talking a bit about the companies that are going to do well in this environment, being capital-light. Or AI, I should say. The AI era is capital-light. And it’s more not physical capital, but others. Maybe you could just tell us how you think it’s going to shape the markets in the next little while. And I know you’ve got a few names just to demonstrate how they may be thinking about AI.
Bill Priest: Well, it isn’t that companies are necessarily capital-light to begin with, and I’ll give you an example. John Deere (DE) is probably the largest farm equipment company in the world, and we think of them as having spreaders and combines and big tractors and whatnot. But actually, they have been a leader in adding data scientists and technicians to their company. The way to think of John Deere today is a software company.
Essentially, what they do — they have software as a service, but it comes through the iron, so to speak, the physical manifestation of that — the spreader, the tractor, or whatnot. But the fact that they can go down a row of corn or a row of things and basically be able to spew out something like fertilizer or something to kill weeds — it’s amazing. And their goal is actually to be able to produce farming with virtually nobody running the farm. It’s a very ambitious goal.
We have one client who owns a great deal of land out in both Iowa and South Dakota, and he’s tripled his acreage, and he hasn’t done anything with his employees. It’s all been the ability to use, in this case, largely, John Deere’s equipment and the technology that’s now built into the applications that Deere offers.
Another example would be Microsoft (MSFT). Now, that seems quite obvious, but basically, AI from Microsoft could easily provide an acceleration in their growth rate, perhaps as much as 10% to 30% through Azure and what’s called Copilot. The Microsoft 365 Copilot — it’s a “$30 per user per month” add-on to Office. In Word, it creates a first draft. In PowerPoint, it can start a presentation based on your other documents. In Excel, it creates charts and analyzes data for you. It’s the copilot concept of man and machine working together at the office. Microsoft applications are the obvious place to put AI copilots. That would be a second company.
And another one in the healthcare field, which Kevin was discussing, would be UnitedHealth (UNH). AI has lots of applications in these administrative and somewhat tedious businesses, like health insurance. And UnitedHealth Group is a very well-known managed-care company. Coding, for example, is basically classifying and measuring patient illnesses. It’s critical to how health care is priced and financed. And you may have seen headlines about RADV audits, which are government audits of coding practices at managed-care companies that could impact billions of dollars in profit. And improving coding accuracy is an excellent AI application. In the near term, UnitedHealth is looking to use generative AI to help more efficiently write medical appeal letters and prove payment integrity models, detect fraud and abuse, and really answer basic questions in their call centers. Longer term, AI can create significant general and administrative savings at UnitedHealth.
They have, for example, 260,000 employees that are not clinicians. These are people that do much more administrative tasks than anything else. And if you think about the typical salary for medical administration, the number of employees, the savings potential here is several billion dollars a year.
So it really is — most companies are spending an enormous amount of time finding applications for AI, and it will do one of two things. It’s either going to improve margins for them, or it’s going to improve asset turnover. In other words, they’ll be able to generate the same level of revenues with lower cost of labor and physical costs as well.
Kim Parlee: It’s funny. You’re going through these examples, and, Bill, each one of them is jaw-dropping in what it can do. So it’s incredible just to think of the mass adoption and what it could mean. I’ve only got about a minute, but I wanted to ask you, when you think about geographies, country-level allocation, what does the AI thesis tell you there that you might want to tilt more towards?
Bill Priest: Well, I think the developed countries, for the most part, will see the initial benefits of this. So it’s going to be the West — perhaps in China as well — but you’re going to see a lot of that come in the West, I think, initially. You’re going to see companies that are already involved in using technology that get the idea that there are two big questions in running a company today when it comes to allocating capital.
One is, what criteria do you use to allocate capital? But the second one is going to be, how do I run my business in an age that is evolving very, very rapidly? How do I manage my business in a digital age? Do I have a business strategy, and can I use AI to enhance my business strategy? I don’t think there’s a company in the world that isn’t thinking of it in those terms.
Editor’s Note: The summary bullets for this article were chosen by Seeking Alpha editors.