How Will AI Impact Investing?

Welcome to "The Informed Investor," where we break down the latest financial headlines, bringing in research and insights to help you separate the news from the noise. Welcome everybody, and thank you for joining "The Informed Investor," a show brought to you by Dimensional Fund Advisors, a $900 Billion global asset manager. And Today's Topic, is artificial intelligence, and where does it fit in my investment portfolio? I'm Mark Gochnour. I'm joined today by Jake DeKinder and Dr. Wes Crill. And Dr. Wes Crill this is kind of right up your alley a little bit, isn't it? Going back to your PhD. It kind of is amazingly enough. You know, when I was, probably my second year of grad school, we were doing all these like large scale molecular dynamic simulations, basically simulating the motion of a whole bunch of particles, and it's an exhaustive computing task. And so we were actually evaluating some, what we called graphical processing units, these huge computer chips that could process lots of things in parallel. And oddly enough, they were made by NVIDIA. Too expensive for us, for grad students, but it would've been really beneficial to our research at the time. How much the stock did you buy in NVIDIA at that time? I was a little short on cash as a grad student, so I liked to buy none. If you only knew, only knew at that time. It would've been great. All right, I'm gonna read a couple headlines. Jake, then I'll come to you and get your thoughts on these. Alright, the first one, the AI bull market still has another two to four years left. And interestingly enough, this next headline was written the same day. AI stocks are in a bubble. Why are so many investors refusing to believe it? And then the last one here, which is fantastic, is it time to sell your AI stocks? I mean, classic market timings, classic market timing on that last one there, right? But then also the, I was gonna say the two to four year. I mean, just, I love the precision around there of like, it's, you know, it's like a two to four year window. You're like, I, you know, if I'm gonna be a market timer, I maybe need a little bit more help than that. Well we always talk about that, some of the prognostications are oddly specific when it comes to those analysts. I actually have, not a title but an excerpt. I get really jealous when you guys are already reading the headlines. So I wanted to bring this one up because I think maybe it plays into the conversation today. Now this is from James Cameron. Of course everyone knows James Cameron, famous director. So he was asked a question about writing science fiction stuff, like what was he currently working on. He said, "I'm in a point now where I have a hard time writing science fiction. I've been unable to get started on 'Terminator 7,'" Lord knows we need another one of those, "because I don't know what to say that won't be overtaken by real events." Essentially, we're already living in sci-fi. It's hard to write fiction anymore. Man, I wanna come back to, that's crazy, that headline, right? I mean there's maybe a little bit of of truth to that. Absolutely. That is evolving. Very, very fascinating. There's a lot of stuff that's coming out that's surprising people. I actually want to come back to your example from being a grad student because everybody's talking about this idea of AI being something new, right? And I don't think you got your PhD a year or two ago. I know you've worked a Dimensional about 15 years, Feels like just yesterday, but it was a little while ago. Right? And actually, we've done an exercise where we've gone back and looked at different headlines. In fact, we had a webcast a number of years ago, or maybe a year, I don't remember exactly. But anyways, and we looked at these headlines that had come out and one was from 2013, 2014, 2015, all around AI. And then we found that one. Do you remember? It said AI has reached the golden age. 50 years, AI had been around? That headline was in 2006. 2006, Yeah. Yep. Yeah, absolutely. So it's been around a while. And I just think about AI too, 'cause it feels like to me, hey, what about AI? And AI, sort of just everything, you know, and it's almost as if everything falls under that AI. And I guess I'll get your thoughts on this. I'll start with you, Wes, and, the way I think about AI, to me, there's a couple of different areas there. One is it's just sort of natural evolution of technology. You know, you've been able to do some things a little bit faster, a little bit quicker, a little more, more efficient. So for example, I can go to ChatGPT, do a quick search. It's much faster than if I went to one of the search engines and just had to scroll down everything. Or I can go to my phone, I can pay a bill in a matter of 10 seconds. You know, it's not necessarily new. I could have written a check or gone to the bank, but it's just so much faster. So to me, part of it is just that normal advance of technology from efficiency. The other part of it is, man, there's some really new, really interesting innovative stuff going on here that it's gonna be fascinating to see what unfolds here, when you think about bringing together a whole bunch of data out there in ways we just hadn't quite thought about connecting it, but you know, we'll talk about AI here, but how do you think about AI and a lot of the questions you get from folks when you're out there meeting with clients. Yeah, and I should say I'm far from an expert on AI type algorithm development. But I think my experience as a grad student provides a little bit of context for what I think is the big breakthrough here, which is being able to process a lot of information simultaneously in parallel, the operational efficiency of these, what they call large language models of summarizing things from anywhere on the internet. When you interview query ChatGPT, that really comes from their ability to organize and access very disparate pieces of information and bring them together almost like a neural network for the human brain can process things. And so I think that element of it is a pretty big breakthrough. Certainly part of the reason why we had a recent, you know, you mentioned the stuff, the idea has been around for a long time. Part of the reason for the more recent breakthroughs is the computing power. And I think that's really where you're seeing a lot of the operational advances. Alright, Jake, you meet with a lot of clients, both financial professionals and investors out there. What are some questions you're getting from investors that we're gonna tackle here today? We're out meeting with investors all the time. We're meeting with financial professionals all the time. Three big ones that I think pop to the top of the list of what I hear, if I can kind of group 'em together. One is that, is it somehow gonna change how markets function? Meaning how buyers and sellers come together to transact into set prices. The other, the second one is, does it somehow give certain people a competitive advantage in the marketplace? Allow them to pick stocks, better time markets, as we were referencing all of those things better. And then the third one, and this one maybe comes a little bit more from the investor side is, either, am I missing out or how do I play this as an investor, right? I've heard about NVIDIA, I've heard about all of these companies, a lot of 'em have had incredibly strong returns. Am I somehow missing out on all of this stuff that's being discussed around AI and how do I get me some of that? Well maybe we start with that one. Kind of run with that a little bit on the context of how our company's using AI. Like am I getting some exposure through various companies or do I have to go directly to an AI oriented company to get that? I mean, depending on how they're using it, there's probably not gonna be any companies that won't eventually be touched by the idea of artificial intelligence. You know, we mentioned operational efficiencies, being able to, I dunno, service your customers more effectively or even some of the cool ones. And we've talked about some of the companies that we observe are in these artificial intelligence focused strategies out there. And some of 'em are that really sci-fi sounding right? Companies like Caterpillar, companies like Honeywell, I don't think of those as being part of the plot for the next "Terminator" movie, but they are, you know, in their own way using artificial intelligence tools to grow their business and further their growth. Well it kinda goes back to it's been around a while. I remember you were joking as we were talking about this episode about Clippy. Yeah. That was my first experience with AI, right? So you start typing in Microsoft Word in the late '90s and you say, "Dear John," put a comma after it hit return and then it would pop up the enterprising paperclip known as Clippy and he'd say, "It looks like you're trying to write a letter. Would you like some help with formatting?" And that is a very rudimentary version of what we now think of as artificial intelligence deployment, which is recognizing a pattern. Now I can recognize a pattern because it's been fit to previous circumstances or data and then making a prediction or offering guidance or summarization from there. When, well, sorry, go ahead Jake. Well no, I was just gonna say, you know, how are companies using it? I think at least at a high level, I think every company's figuring out how they can use AI to make their businesses more efficient, to reach new markets, to segment their customers better, to learn more from the massive amounts of data that a lot of these companies are already, I mean, you had years where people were sitting on data and didn't know what to do with it inside a companies, right? So then I think about it from an investment standpoint and you're like, well how do I play that? And you're like, well if every company's out there trying to figure out how they use AI to make their operations more efficient, to improve cash flows, better profit margins, all of that, the way I play it is I invest in all of these companies. It's a broadly diversified portfolio. And the one that we always reference back to is, you know, something like the barcode, you know, when the barcode was invented, was it just the company that invented the barcode that benefited from that and the people that invested in that company? Or do you think it also probably helped a lot of the other companies that figure out how to incorporate the new technology into their operations, right? All of the retailers, all of the shippers, all the ways that you can use that technology. And I look at it from a very similar standpoint of where we're at right now. So I don't have any FOMO that it's somehow missing out as long as I think about it in terms of broad exposure to a lot of companies. When you're getting that, if you just look at the holdings of these AI focused ETFs, we did a little study with the top five largest AI based ETFs and if you look at their holdings, over 40% of that was overlapping with the market portfolio. So if you have a broadly diversified portfolio, likely are holding a lot of that stuff that people believe is ultimately gonna be a beneficiary of AI. When you mentioned one of the names earlier, I think Caterpillar. Yeah. I think that's one of the holdings in some of those ETFs you just alluded to in those top five ETFs around AI. And you think, well Caterpillar, or, I think, Deere companies in there, think, well how can that be? It goes back to some of these examples where this technology's been around a while and I think we've all been to our Australian office down there doing some work over time. And in Western Australia you have these massive mines and I think all these, you know, 300 ton dump trucks, I think it was the late 2000, like 2008'sh is when they just started running 24/7 with no drivers or it was just all through GPS, different radars, sensors to where they're going. And I think since then it's been much safer than having a human driver. So to me that's an early form of, again, some of this AI that's been driving this. And to your point about efficiencies that we all benefit from, somewhere in a globally different side portfolio, all the companies we own, it's finding its way into all these different companies. You can really take comfort of that of any time a new technology comes out, right? I mean if you go back through time, new technologies come out, companies adopted, they figure out, and by the way, those that don't can very well go outta business. Which is why again, if you go from diversification to concentration, you can run into some issues there. 'cause you're basically betting on, will this individual company adapt to a changing world which is always taking place. There's a bit like the dot com revolution, right? I mean how often do you guys buy anything on pets.com? I do not, not on pets.com. No. So I think that was one of the challenges back then was okay, we think that the internet revolution is gonna be, it's gonna impact every company out there in some way, shape, or form. I just don't know who's gonna be the winner and who's gonna be the loser. NVIDIA looks like they were an early winner, but we don't know whether companies are gonna be able to leverage this better than their competitors and truly outperform. By the way, do mascot for pets.com, do you remember it? The sock puppet? Talking sock puppet. Always a good investment. How could we forget? That's right. Let's hit your other question, Jake, does AI give you some competitive advantage when it comes to investing to somehow we create this crystal ball perhaps that's been missing prior to, I guess the ability to use AI, in whatever shape or form about predict future prices and stock or bonds? Let's go start with you, Wes. Yeah. Take that one Wes. Yeah. Yeah. I mean, okay, what it would have to do to give you an advantage over other investors is make the information retrieval superior in some way. Either find information that's alluded to capture the rest of the market or allow you to access it more quickly than anyone else out there. You know, we see in the data, I mean last year in equity markets alone, there's about $800 billion worth of trade volume on a daily basis. That is a lot of activity looking for information that is not market prices, that is really what keeps markets in equilibrium and makes it very challenging to out guess markets. And so you think, okay, well maybe these AI algorithms can find something out there a little more quickly and get an edge. But you know, you look at the historical data around the paucity of managers who were able to outperform the stock market, there's not a whole lot of evidence they're missing out on anything because so few of them could add value over index benchmarks and you know, if they do find something, let's say they find a new avenue to quickly access something that everyone else is missing out on, that's kind of a one time thing, right? Once you discover that avenue of information retrieval, then it becomes part of the market's information set and it's no longer usable for something like alpha out there. And by that you mean once it's starting to be traded upon, then it's out there, people are gonna observe- Traded again prices. Maybe they make money on a one time event, but then at that point that's not a continuous basis for outperforming markets. Yeah, there's probably observable somewhere in the marketplace perhaps. Well you go back to, I mean I'll just talk about, sort of the advancement of technology through time. I mean the first studies on manager outperformance versus sort of a passive buy and hold benchmark, go back to the mid to late 1960s, right? And since then we've had a lot of technological advancements and yet when you still look at the data through the decades and we've run plenty of these studies, other companies have run plenty of studies, lots of academics, you know, you don't really see too many professional managers outperform a passive sort of benchmark than what you would expect by chance. So yes, it is different. Yes it is new. Yes, the computing power is stronger. But again, I think probably all along the way people were making a similar argument and what the evidence shows is we haven't necessarily seen anybody really get an edge. You remind me of a story, Dan Wheeler. Yeah. A long time colleague who started the advisor business here at Dimensional. This was around the tech boom in the late nineties. And he was having breakfast and he heard somebody over talking in the booth next to him and somebody was just telling their buddies like, all I need is a faster computer and my trades will get in faster, they'll make all this money in my day trading approach. Because whatever he was doing wasn't working. People have been making this argument forever, right? But the problem is, is that the future is unpredictable, right? And so even with the best models, even with the best data, the ability to extract it, to piece things together, to do all of that, you just, you don't know what's gonna happen in the future. And you don't necessarily know how markets will react to future events. And we've got a lot of evidence of that. You know, we cite 2020 and I don't think a lot of people, if you came to 'em in January of 2020 and said here's what's gonna happen in late Feb and March, do you think the market's gonna be up for the rest of the year? Not a lot of people you know would've said that. And there's a lot of these examples through time and that's hard. You can't predict the future and you can't predict how markets will react to future events. That's a good point about not being able to predict markets because you know, we know based on the automated vehicles at Waymo's, we're starting to see more and more in Austin. We know they can predict when to stop at an intersection because they look up and they see a stop sign. And they know that's their indication to slow down and stop. But then, you know, you think about why they can actually make that determination and it's time to stop. Well, because they know what a stop sign looks like. They know what a stop sign looks like. 'cause it doesn't change from one day to the next. It's not octagonal today and hexagonal tomorrow. And that's the challenge with trying to predict market movements is if you're gonna try and fit it to data, well stock and bond returns are incredibly noisy and you know, the correlations across different securities are not long lasting. There's no stability in the patterns. You can't just fit it to a data set and say, oh now I'm gonna make a better prediction of the future because it's likely to be garbage in, garbage out, in terms of the noise level. Well, and it's, not saying here that, you know, somebody using AI to try to manage money is necessarily gonna have a lot of trading, but it's a potential, right? I got a lot of data, it's coming in, there's a lot of noise in markets, right? I do a lot of trading, now my cost could be higher. I gotta think about taxable events. So there's a lot that you really have to think about, once you go down that path of here's how I'm gonna use massive amounts of data to do a whole lot of movement inside of how I'm managing money. When you talk about noise, are you just talking about how fast and to the degree prices change immediately in stocks and bonds? Yeah. Is that what you mean by noise? Yeah, I mean stock and bond returns are incredibly volatile, especially on an individual base. You look at the overall markets return to get diversification across these securities, but on an individual stock and bond basis, these things are very volatile. And so trying to, you know, anytime you're trying to get a model that does a good job of predicting things in the future, it's really a signal to noise ratio calculation you're doing well, no matter how strong your signal is, if it's completely drowned out by noise, it's not gonna be additive to your process. And I think that's where, you know, look, these artificial intelligence algorithms are gonna do lots of potentially great things or informative things, but doing better than the stock market I think is a really high bar to set. And as an individual investor, you have to ask yourself, do you really have to play that game to have a good investment experience, right? And we say that all the time. I mean, yeah, do you want to have higher returns? Do you wanna do all of that? Sure. The downside could be you don't get the higher returns, it could be more costly, it's a tax efficient way to go versus are you saying I want to have a sensible long-term approach that helps me reach my goals and not necessarily stress about a lot of the noise maybe that I'm hearing out there. I'm gonna go back to what you said about sort of the patterns, right, where AI is really good when there's a pattern you're following, but because the prices change so quickly in stocks and bonds, there's no way to really use that pattern to identify the future because the future's always uncertain, right? So I think you summarized that really well there, Jake, about you can have a very successful investment experience without having to try to time the market or predict what's gonna happen in the future. Yet you're still getting the benefits of all this AI out there through these companies that are using it globally in creating more efficient operations. You also wanna think about what expectations for the future are already priced in. So the companies that we believe are gonna benefit greatly from the expansion of artificial intelligence algorithms, it's probably already priced in. Like if people think this company is gonna do particularly well, they're gonna be willing to pay a higher price. Honestly, when we look at the data around a lot of these AI focused ETFs, a few of them have done better than the broad market, but only maybe one or two have done better than the tech sector. And that's a useful benchmark by the way. 'cause a lot of these early movers are technology companies. But then when control for that element of the performance in markets, they don't seem to be delivering anything special because, again, the extent to which that's already expected, it's gonna be priced in, you're not gonna generate abnormal returns from it. And we know tech sector's done well recently, that just is what it is. But so almost by default, companies that sort of operate in that space, they've probably done reasonably well. But it's an interesting data point you cite around the ETFs that are specifically focused on that. I liked what you said too about what we know today's already reflected in prices and then whatever's going to be revealed in the future at that point it'll be revealed in prices. Well, yeah, yeah. Market moves in, right? Yeah, and if you knew exactly which companies were gonna get unexpectedly good news in the future, yeah you would absolutely wanna load up on those. Well, and like any of those new technologies, we know there's gonna be positive elements of it and there's gonna be negative elements of it and we don't need to get into a societal discussion as well. But that also can sort of translate into what that might mean for you as an investor. If you go down too far down that path of concentrating on, I wanna load up on XYZ technology because we don't know what it's gonna look like in the future. Well I might have some kinks to iron out too. I know that we were talking about this before, but one of the liquor stores, it's right down the street from here, their vendor uses artificial intelligence to sort of, I guess quick check all of the orders that come in and if they're outside of cluster as they describe it, so something that looks abnormal for that time of year, that particular brand and they'll reject it. And so they were having some teething issues in terms of properly stocking their store because the AI algorithm kept saying, nope, you don't get that many bottles of it. And if I know where you're going with that, particularly that weekend, because there was a University of Texas game and their big annual sale was around the same weekend. So I imagine there was a huge cluster. Yeah, well that might have been a good example of an outlier where the AI algorithm hadn't even considered the possibility that there was going to be a sale at this liquor store and YouTube was gonna have a home football game. And you can't be running outta Tito's when you have a home football game in Austin. Seriously, man, That's, I mean unless you teach then start playing some better football man, for God's sakes. Exactly. Maybe I need more Tito's or less, I don't know how it shakes. And the tough ones are coming too. That's right. Yeah, yeah. We need more Tito's for the October games. Alright, so I go back to those three questions you brought up the very beginning. I guess the answer to me, what I'm hearing, is global diversification works, I should say global diversification works 'cause you're gonna get exposure to all of this AI and there's nothing we've seen yet or really would anticipate that can somehow identify the future. Anything else you'd add to that? You put it succinctly. It's one of my favorite quotes by the way, from "Robin Hood: Men and Tights." The guy says that. And no one else at the table has any idea what he was talking about. But I think it's a really nice way to describe the way you summarize things. I like the Simpsons one where they're on the ship and he says, "I'm a man of few words. Any questions?" We're have to turn this into just a movie and TV show review podcast at some point And with that one, thank you for joining "The Informed Investor" today. Be sure to hit the subscribe button and have a fantastic rest of the day.