Show Transcript
While artificial intelligence may have a major impact on jobs and productivity, is the recent run higher in AI related stocks showing signs of a bubble? Greg Bonnell discusses with Kevin Hebner, Global Investment Strategist with TD Epoch.
Print Transcript
[AUDIO LOGO]
While there is a lot of excitement around the development of artificial intelligence technology, also concerned about how disruptive it could be to the labor market and whether assets in the space are in a bubble. Lots to discuss here. Joining us is Kevin Hebner, Global Investment Strategist with TD Epoch.
Kevin, great to have you on the program. Big topic. I know there's a lot of interest out there. We have so many things to go through. Let's go through the disruptions. I think this is what people are worried about when they hear about AI. How disruptive could it be to our lives, to the labor market?
Hi, Greg. I think that's true. And certainly, when you look at polls, more people are worried than excited about AI because they are concerned about what will happen with their jobs going forward. So in our view, there's going to be about 60% of jobs will be significantly affected by AI. But in fact, that's pretty normal.
If you think right now, 60% of occupations that exist today did not exist in 1950. So the labor market is always very dynamic and lots of change. But the way we come up with these numbers is the US Department of Labor divides the US labor market into 1,000 different occupations. Then each occupation they break into 15 to 20 different tasks. So you have 18,000 tasks overall describing the US labor market.
Then you look at which tasks can be impacted by existing AI-- so not AI as it will be in five years or 10 years but as exists today. And then you get these numbers-- 60% being significantly affected, and for, say, 50% of jobs, over 20% of tasks will be affected. So there will be a lot of jobs being changed and certainly more jobs being changed than displaced by AI. But it means a lot of change going forward, and I think something which we all have to respond to, and I think responding to positively.
Yeah, because what we hear about it from people who are proponents of it, they say it's going to take away the drudgery work, right? And I think of some people I talk about in financial services say, I spend a lot of time charting this or putting all these numbers together. The AI does it for me, frees me up to do other things. I think where people get concerned is that they're worried that I've got one of the drudgery jobs, and that job will just be gone altogether.
I think for most of us-- and it's interesting. If we were having this conversation 18 months ago, we would all have agreed, well, AI is first going to affect blue-collar physical occupations, then white collar, and then creative. But with generative AI, OpenAI and such type of models, this has been totally flipped. And now it's creative jobs, then white-collar jobs, and then blue collar jobs that are being the most impacted.
And so I think it's important to see that, as this evolves, how we view things is going to be changing a lot. But, say, I'm in the white collar sector, and for a lot of us, AI, it's augmenting what we can do. So the Excel work I do, downloading data, regressions, creating charts, creating pitch books and things-- AI will certainly be improving my productivity, say, by 20%.
So I've got a long to-do list, so that's going to help me get through more things. And I think, broadly, for white collar workers, that's true. Creative workers, it's even more so. People who write, it's going to increase your productivity by 40%. Marketing copy is, in fact, the number one use of generative AI so far.
People who create digital economy content animation, if you saw Sola, it's a tool created by OpenAI. It immediately increases the productivity of people in the automation business-- video games and so forth-- by about 90%. So it's pretty dramatic. And so we will see a lot more content, hopefully not just mid content, but excellent, really engaging content coming forward.
You gave us some concrete examples there of productivity, whether it's in financial services, the creative arts. What does that mean on a national level when we start measuring an economy's productivity, when we start measuring GDP output? Could AI be the thing that sort of starts kickstarting productivity where, perhaps, we have lagged for a while?
Yeah, we think that it's going to increase productivity relative to baseline by 10% to 15%. And we come up with that number by looking at what's happened historically, say, with electricity, or computers, or the internet. So it's been similar to that. You can also do it by bottom-up-- what we were mentioning before with the 18,000 tasks and looking at the impact on productivity and output.
But either way, we come up with a number-- say, over the next decade or so-- about 15%. And that's a very big number. That's similar to, say, the internet boom that it experienced in the late 1990s. So it is a big deal.
Now, when we talk about productivity and we talk about it happening across a broad swath of society in different industries, obviously, as investors, we start to think about companies that are more productive would probably make better use of their inputs and then see better profits. I mean, is this the flow-through, then, to the investment community where more productive society, more productive corporations actually produce healthier profits?
Well, certainly, if you have higher productivity, higher growth, you're going to have higher topline growth going forward for companies. But we're also seeing right now, at least for companies who are directly benefiting from the buildup of the infrastructure for platforms, the picks and shovels, if you will-- very dramatic increase in margins, return on invested capital, and generating pretty phenomenal free cash flow. That's for a small number of companies in the overall market. So we're seeing very concentrated returns as we build out the infrastructure for AI.
I think about the chipmakers. Obviously, in the early innings of our discussions about AI, and we're about a year into the real proper discussion on a broader base in society. The chipmakers have benefited. There's got to be a bit more here, I imagine, as we go forward, than just playing it through chipmakers.
I think, right now, we're building out the infrastructure-- so for the platforms, different places in semis. But right now, we don't have very much in terms of applications. And typically, when you have a new disruptive technology, a general purpose technology, we like to say, it takes a long time for it to diffuse across the economy.
Often, it can be 20 to 30 years to get a 50% diffusion rate. So AI might be a little bit faster because it happens primarily in the world of bits. But it's still a very slow process. So right now, we've got this boom as we build out the infrastructure. But then going forward, we have to get to the point where, where's the beef?
Where are the killer apps that help companies and households going forward? And I think that's going to take longer than many people think. So on our view, we're going to keep enjoying the type of returns we've had over the last couple of quarters. But at some point, there's going to be this chasm where the headlines are going to be saying, where's the beef? Where are the killer apps? How is this really helping companies and households? And I think that takes a lot longer than the investment community is currently expecting.
Now, the investment community, if it's going to take longer than they're currently expecting in terms of us feeling our daily lives changed by these killer apps, as you call them, from AI-- we've seen a big runup in those chip names. Again, NVIDIA has been a poster child here. It has people saying the word "bubble." If you take a look what's happening AI right now in the market runup, are we in a bubble?
Yeah. So there's three reasons to think we might be in a bubble, say, similar to where we were in the 1990s. One reason is the concentration of returns. And in fact, since 1920 there's, only been two other times when returns have been so concentrated in a small number of names-- the late 1990s and also the late 1920s, and neither of those ended up terribly well.
The second reason is valuations. Valuations are stretched-- not as much as they were in the 1990s but, clearly, stretched beyond norms. And the third reason, I think, is the quite euphoric tone of many pundits in a lot of commentary about AI. So that's three reasons to think we might be in a bubble.
But what is different this time is, in the 1990s, companies were burning through a lot of cash. And ultimately what burst the 1990s bubbles were the headlines and Barron's and things saying, oh, my god, we're running out of cash. They're all burning money. And this thing can't run very much longer.
Right now, we've got enormous free cash flow generation, great margins, great return on invested capital, at least from Mag Seven and related names. So that makes it different. And we would say it means that now is more like 1997 than 1999. [AUDIO LOGO]
[MUSIC PLAYING]
While there is a lot of excitement around the development of artificial intelligence technology, also concerned about how disruptive it could be to the labor market and whether assets in the space are in a bubble. Lots to discuss here. Joining us is Kevin Hebner, Global Investment Strategist with TD Epoch.
Kevin, great to have you on the program. Big topic. I know there's a lot of interest out there. We have so many things to go through. Let's go through the disruptions. I think this is what people are worried about when they hear about AI. How disruptive could it be to our lives, to the labor market?
Hi, Greg. I think that's true. And certainly, when you look at polls, more people are worried than excited about AI because they are concerned about what will happen with their jobs going forward. So in our view, there's going to be about 60% of jobs will be significantly affected by AI. But in fact, that's pretty normal.
If you think right now, 60% of occupations that exist today did not exist in 1950. So the labor market is always very dynamic and lots of change. But the way we come up with these numbers is the US Department of Labor divides the US labor market into 1,000 different occupations. Then each occupation they break into 15 to 20 different tasks. So you have 18,000 tasks overall describing the US labor market.
Then you look at which tasks can be impacted by existing AI-- so not AI as it will be in five years or 10 years but as exists today. And then you get these numbers-- 60% being significantly affected, and for, say, 50% of jobs, over 20% of tasks will be affected. So there will be a lot of jobs being changed and certainly more jobs being changed than displaced by AI. But it means a lot of change going forward, and I think something which we all have to respond to, and I think responding to positively.
Yeah, because what we hear about it from people who are proponents of it, they say it's going to take away the drudgery work, right? And I think of some people I talk about in financial services say, I spend a lot of time charting this or putting all these numbers together. The AI does it for me, frees me up to do other things. I think where people get concerned is that they're worried that I've got one of the drudgery jobs, and that job will just be gone altogether.
I think for most of us-- and it's interesting. If we were having this conversation 18 months ago, we would all have agreed, well, AI is first going to affect blue-collar physical occupations, then white collar, and then creative. But with generative AI, OpenAI and such type of models, this has been totally flipped. And now it's creative jobs, then white-collar jobs, and then blue collar jobs that are being the most impacted.
And so I think it's important to see that, as this evolves, how we view things is going to be changing a lot. But, say, I'm in the white collar sector, and for a lot of us, AI, it's augmenting what we can do. So the Excel work I do, downloading data, regressions, creating charts, creating pitch books and things-- AI will certainly be improving my productivity, say, by 20%.
So I've got a long to-do list, so that's going to help me get through more things. And I think, broadly, for white collar workers, that's true. Creative workers, it's even more so. People who write, it's going to increase your productivity by 40%. Marketing copy is, in fact, the number one use of generative AI so far.
People who create digital economy content animation, if you saw Sola, it's a tool created by OpenAI. It immediately increases the productivity of people in the automation business-- video games and so forth-- by about 90%. So it's pretty dramatic. And so we will see a lot more content, hopefully not just mid content, but excellent, really engaging content coming forward.
You gave us some concrete examples there of productivity, whether it's in financial services, the creative arts. What does that mean on a national level when we start measuring an economy's productivity, when we start measuring GDP output? Could AI be the thing that sort of starts kickstarting productivity where, perhaps, we have lagged for a while?
Yeah, we think that it's going to increase productivity relative to baseline by 10% to 15%. And we come up with that number by looking at what's happened historically, say, with electricity, or computers, or the internet. So it's been similar to that. You can also do it by bottom-up-- what we were mentioning before with the 18,000 tasks and looking at the impact on productivity and output.
But either way, we come up with a number-- say, over the next decade or so-- about 15%. And that's a very big number. That's similar to, say, the internet boom that it experienced in the late 1990s. So it is a big deal.
Now, when we talk about productivity and we talk about it happening across a broad swath of society in different industries, obviously, as investors, we start to think about companies that are more productive would probably make better use of their inputs and then see better profits. I mean, is this the flow-through, then, to the investment community where more productive society, more productive corporations actually produce healthier profits?
Well, certainly, if you have higher productivity, higher growth, you're going to have higher topline growth going forward for companies. But we're also seeing right now, at least for companies who are directly benefiting from the buildup of the infrastructure for platforms, the picks and shovels, if you will-- very dramatic increase in margins, return on invested capital, and generating pretty phenomenal free cash flow. That's for a small number of companies in the overall market. So we're seeing very concentrated returns as we build out the infrastructure for AI.
I think about the chipmakers. Obviously, in the early innings of our discussions about AI, and we're about a year into the real proper discussion on a broader base in society. The chipmakers have benefited. There's got to be a bit more here, I imagine, as we go forward, than just playing it through chipmakers.
I think, right now, we're building out the infrastructure-- so for the platforms, different places in semis. But right now, we don't have very much in terms of applications. And typically, when you have a new disruptive technology, a general purpose technology, we like to say, it takes a long time for it to diffuse across the economy.
Often, it can be 20 to 30 years to get a 50% diffusion rate. So AI might be a little bit faster because it happens primarily in the world of bits. But it's still a very slow process. So right now, we've got this boom as we build out the infrastructure. But then going forward, we have to get to the point where, where's the beef?
Where are the killer apps that help companies and households going forward? And I think that's going to take longer than many people think. So on our view, we're going to keep enjoying the type of returns we've had over the last couple of quarters. But at some point, there's going to be this chasm where the headlines are going to be saying, where's the beef? Where are the killer apps? How is this really helping companies and households? And I think that takes a lot longer than the investment community is currently expecting.
Now, the investment community, if it's going to take longer than they're currently expecting in terms of us feeling our daily lives changed by these killer apps, as you call them, from AI-- we've seen a big runup in those chip names. Again, NVIDIA has been a poster child here. It has people saying the word "bubble." If you take a look what's happening AI right now in the market runup, are we in a bubble?
Yeah. So there's three reasons to think we might be in a bubble, say, similar to where we were in the 1990s. One reason is the concentration of returns. And in fact, since 1920 there's, only been two other times when returns have been so concentrated in a small number of names-- the late 1990s and also the late 1920s, and neither of those ended up terribly well.
The second reason is valuations. Valuations are stretched-- not as much as they were in the 1990s but, clearly, stretched beyond norms. And the third reason, I think, is the quite euphoric tone of many pundits in a lot of commentary about AI. So that's three reasons to think we might be in a bubble.
But what is different this time is, in the 1990s, companies were burning through a lot of cash. And ultimately what burst the 1990s bubbles were the headlines and Barron's and things saying, oh, my god, we're running out of cash. They're all burning money. And this thing can't run very much longer.
Right now, we've got enormous free cash flow generation, great margins, great return on invested capital, at least from Mag Seven and related names. So that makes it different. And we would say it means that now is more like 1997 than 1999. [AUDIO LOGO]
[MUSIC PLAYING]
