It is August 28th. of 2024. Why am I being so specific? Because late after the market closes today, NVIDIA is going to report its earnings.
The entire market is waiting to see what those numbers will be. As AI has taken over the market and it's driving narrative, it's worth remembering how recent this talk is. It was in November of 2022 that OpenAI introduced ChatGPT.
Now ChatGPT is low-tech AI but let's face it, EPPT is what opened the door for everybody to recognize how much AI could be not just a market mover, but how much it could change the way we work and live. In a post last year, I looked at the AI effect on businesses. I argued, like other big revolutionary changes, and I do think AI is a revolutionary change, it will create some big winners.
Right now, we've anointed NVIDIA, Microsoft as those big winners, but who knows who the final winners will be. But I also argue that you're going to see a lot of wannabes, companies that use AI as a guise to push up their values, even though they get little material benefit from AI. And you're going to get lots of losers. You might say, why? Well, disruption always has a cost, right?
I mean, we think about the good side of disruption, new companies coming up, but those new companies often come at the expense of older ones. You're going to have some winners. You're going to have a whole host of wannabes and quite a few losers.
Now, we can talk about business disruption and perhaps I will do a session on that later. But in this session, I want to focus on disruption at the personal level. What am I talking about?
You probably read about the fact that AI is coming for you. Just a couple of weeks ago, I was reading an article in the New York Times about how many jobs would be replaced by AI entities. Marc Andreessen is convinced that all our jobs are going to disappear in a few weeks. a decade because AI is going to take it over.
And I've been thinking about this because this is a question that I can no longer avoid and here's why. As you may know, I teach my classes in the spring semester of each year. So this year I taught my classes between February and May and in about the 11th to 12th week of the semester I got a call from a friend of mine, Vasanthar, who teaches machine learning at NYU Stern School of Business. Now, Vasant has forgotten more about AI than I will ever learn in my lifetime.
He's an expert on AI. So I was interested in why he was calling me. He said, Aswath, we've designed a demodern bot. I said, a what?
He said, a demodern bot. He said, we've created an AI entity that has read every single one of your blog posts. And I am a verbose writer.
I've written about 2,000 pages worth on my blog. He said, it's read every one of your books. It's looked at every one of your evaluations. It's watched every one of your YouTube videos, which includes every class I've taught.
In effect, since everything I do is in the public domain, the Demodaran bot has not just read it and watched it, but it remembers everything. Remember, computers have a memory that doesn't slip. And I said, so what do you want from me?
He said, we're ready. And I said, ready for what? He said, we're ready to turn the bot loose to value companies. And we'd like to run a contest where the bot will value a company and one of your best students will value a company.
from your class and we're going to see which one does the better job. I don't have the results of that test yet and I will let you know when I do, but I'm terrified of what Vasant will find. If he finds, for instance, that the Demodaran bot does a much better job at valuing companies than my very best students, I'm doomed.
I'm headed for obsolescence. If he finds that the Demodaran bot does a really bad job, then I'm screwed too. And here's why. Everything I do...
everything I write, everything I teach, everything I talk about, is from the perspective of teaching people how to do valuation. If after reading and watching all of those, the demodern bot can't value a company, then what exactly am I teaching? So I'm waiting to see what the answer is and I'm terrified of what we will find.
But that bot has opened up the door as to what exactly I can do and what I should be worrying about. So let's think about gauging the threat to your job to your work from AI. So if you think about AI, AI is not something that was born in November of 2022 when OpenAI announced ChatGPT.
It's a coming together of two forces that have been emerging over the last few decades. The first is increasing computing power in smaller and smaller packages. That smartphone you have is a lot more powerful in terms of computing power than my laptop was or my computer was 40 years ago. So computing power is increasing and getting more compact. The second is data is getting accumulated, not just quantitative data but qualitative data.
Much of that qualitative data has come from us sharing what we do with social media companies in return for convenience. Whether it's Facebook or Google or Apple, the data we share and increasing computing power coming together is what creates AI. As an AI novice, I might be entirely misreading the potential for AI but as I see it there are three dimensions on which you can measure the AI threat. Here's the first one.
Is what you do mechanical or is it intuitive and adaptable? Remember I said AI has deep roots. I am old enough to remember IBM's Deep Blue and Watson and these were computers that had been taught how to play chess.
They'd been fed every chess move ever made in history. And they played against the grandmasters and it turned out that the computers were beating the grandmasters regularly. And why are we surprised?
Chess ultimately is mechanical. The pieces move the way they do and every chess move has already been tried before. And if you feed in every chess game ever played into a computer, it's going to remember every game and every move and what to do next. So AI is going to work really well at things like chess, where it's logical and you can work through every mechanical outcome.
In contrast, much as Elon would like to think that, you know, we're going to start with automated driving next year, two years from now, AI has struggled with automated driving. Not because driving is complicated or the rules are complicated, it's because there are human beings on the surface roads. And human beings behave in unpredictable ways.
AI struggles more when you have unpredictable outcomes, when you have to be adaptable. It doesn't mean it can't do it. It has, it'll struggle more.
So is it mechanical or adaptable? So think about your job. Is the bulk of what you do mechanical? In fact, one of the things I advise people to do is draw a pie chart of what they do every day and how much of what they do is mechanical. The greater the percentage is mechanical, the more you have to worry about being replaced.
An extension of that same idea is what you do rules-based or is it principles-based. In a rules-based discipline, everything is driven by rules. If you're in a rules-based discipline, AI is going to be able to do what you do with ease because it not only can read the rules, it is a rule follower.
Unlike human beings who sometimes take shortcuts, AI will follow every rule. In contrast, if you're principle-based and principles tend to be fewer and you then have to take those principles and come up with judgment calls and analysis on your own way of doing things, AI is going to have a tougher time. Again, not impossible, but a tougher time. So is your work rules-based or is it principle-based? From the perspective valuation, let me tell you what I see.
I don't have much time. for appraisal valuation. I don't do appraisal valuation and I don't think much of it and here's why.
Much of appraisal valuation is done either for courts, for legal structures or for accounting and they're both rule-based. In my view, appraisal valuation, at least as practiced now, is clearly begging to be outsourced by AI. In contrast, if you're doing truly financial valuation, I'm going to argue it's principles-based.
which means I can give you the principles that govern how we measure risk, but I'm not going to give you the rule to use to measure risk, so I want to leave that in your hands. It'll be much more difficult, again not impossible for AI to replace what you do if you're in a principle-based discipline. The third dimension on which I'm going to look at whether AI can replace you is going to be a strange one. Now when we think about designing systems or doing analyses, we make being objective a central theme. In fact, many, again, talking about valuation in specific, many of the valuation groups claim to be objective.
They claim that their appraisers are not biased. Now, I don't want to insult appraisers, but the truth is, appraisals... are biased and it's a bias that leads people to come to you.
Let me explain. A client comes to you to do a valuation. They're not coming in with an open mind saying tell me what I'm worth. Often especially in legal or accounting context they're coming in expecting to see a number and if you give them a number that's different higher or lower They're going to push you to try to move your number to where it should be.
You know what appraisers can do if they're human appraisers? They can bring bias into the process but completely delude themselves into believing their objective. I am constantly amazed at the capacity for delusion where you tell yourself, I'm not biased, I'm not biased, while introducing immense amount of bias into your analysis.
It is true, you can design AI systems to be biased. But here's the problem, you've got to program the bias. bias in. And once you program the bias in, there's no plausible deniability.
As you look at the social media companies coming out and admitting to bias in the way they screen data, they have to admit because it's in the computer code. Bias is something human beings pull off really well because we can be delusional and we can talk out of both sides of our mouth and convince ourselves that we're being objective. Computers have a tough...
time doing that. So as you think about your job, think about is it mechanical or is it intuitive? Is it rules-based or is it principle-based?
And third, how much does your bias play a role in keeping you indispensable? And if you're in one of those jobs where your bias is what's keeping your job going, thank the Lord for it because that's what might keep you protected from AI. So as you respond, as you think about... responding to AI and say, what do I do now? It is something that I don't have a choice of not doing because there is a bot with my name that's coming after what I do.
I've been thinking about this question of what can we do as human beings to make it more difficult to be outsourced by machines or replaced by AI. It's not something that I started thinking about a few months ago when I got that Demodaran bot called, though it kind of made it a higher priority item. It's something I've been thinking about.
about for a long time. As machines have become more powerful, data becomes more ubiquitous. What exactly do we as human beings bring to the process? I don't have the answers but here are some thoughts.
First, over the last century we've moved from being generalists to specialists. I mean take medicine. The general practitioner is not just difficult to find but it's a low person on the totem pole.
Everybody's a specialist. In finance, when finance started as a discipline, people traversed across different aspects of finance. Now, when you go, you know, you talk to somebody who's either an academic or a practitioner in finance, they tend to be specialized. They're specialized in option pricing. They're specialized in trading bonds.
And there's a reason for that, right? If you're a specialist, there's so much knowledge, you can't afford to spread yourself across. There are fewer and fewer people who are comfortable.
across their entire, forget about other disciplines, across their own disciplines, because they're so specialized. And I think we've lost something of value as a consequence. Even in the 40 years that I've been around investment banks, the early 1980s, when you walked into an investment bank, you could talk to people who could talk about equity markets and bond markets.
Forget about markets. They could talk about the theater. They could talk about literature. Now, when you go into...
Most financial service companies, you tend to talk to somebody who's so specialized that they have tunnel vision. So what can you do about it? I don't think I have the answer, but let me give you something that came to me when I was traveling to Florence, visiting Florence on a summer vacation with my family. Now, we were in Florence.
It was a hot day. I think my entire family had gone off to get gelato. And I was sitting on the sidewalk in front of the dorm room. if you've ever been to Florence, it's an amazing structure.
It's one of the large, when it was built, it was one of the largest freestanding domes in the world. That domo was built by an architect called Filippo Brunelleschi. You're saying, so what? Filippo was not an architect.
He was not an engineer. He was not a construction specialist. He was an artist. He taught himself enough engineering, enough architecture. enough construction to be able to build what was then the greatest dome in history.
And he wasn't alone, right? In fact, we coined the term Renaissance man to capture people like him and Leonardo da Vinci, you know, a man who could paint, he could sculpt, he was a scientist, he drew aircraft. Renaissance men or women were people who could do multiple things, who essentially were generalists across disciplines. There are fewer and fewer Renaissance people around. And I think we need more of them.
In fact, the evidence is accumulating that generalists have an advantage over specialists, that generalists bring a perspective that we often lose when we become specialists. So that's the first thing. Okay, second, I'm going to talk about storytelling.
That might strike you as strange as somebody teaches valuation corporate finance, which are numbers driven. Now, starting about a decade ago, I noticed something. that surprised and troubled me.
Now, we have access to more data than ever before in valuation and corporate finance. We have access to more powerful tools than ever before. But the quality of our analysis, in my view, has actually become worse over the last three or four decades. And as I started thinking about why, I think one reason is almost everything we do has become financial modeling.
I've used this term before. We're all Excel ninjas. And we don't... think about what drives our numbers.
We've lost the capacity to connect stories numbers. There is this legend of a left brain and a right brain. Now the right brain controls your story side, your left brain controls logic, or maybe it's a reverse. But we're, you know, more and more of us are acting like we're half-brained. In other words, we're focusing on our strong side and completely ignoring our weak side.
So about almost 10 years ago, in 2017, I wrote a book about connecting stories. numbers. Where I said a good valuation is a bridge between stories and numbers. Every number in your valuation list has a story attached to it and every story that you tell about a company, great management, amazing brand name, has a number that goes with it. Now of course storytelling has become something that Silicon Valley has become really good at.
In fact there are actually places you can go as a founder to learn how to tell stories. But here's the difference. You can tell a big story, you can tell fairy tales.
When you're in business you want bounded stories, stories bounded by reality, what you can do. I think we need to get both sides of our brain going again. Each of us has a strong side.
We're either numbers people or story people. Work on your weak side. That's the second thing you can do because it'll make it more difficult for AI to replace you.
Third, there is a muscle that we don't think of as a muscle called the reasoning muscle. You know what that is? When you ask a question, you try to reason your way to an answer.
Human beings have always had to do it. But in the last couple of decades, I think there's something that's happened that has made it more likely that you will not try to reason your way to an answer. That's something is called the Google search box. I mean let me let's be quite honest, when you have a question now to which you want to know the answer what do you do?
you type in the question into Google search and amazingly you get the answer right away. Think of how much less work it is, how convenient it is. And in an interview I did on teaching a few years ago, I talked about what this is doing to people's reasoning power. Because you can look up an answer, you're not reasoning your way to an answer. I think you're losing that reasoning power.
Now, if you're looking up the answer to a trivia question, it's benign. But I think if you're looking up the answer to what tax rate should I use in valuation, you're missing a chance to try to reason away to an answer. Now you may say, so what if I don't use the muscle? Well, evolution works in magical ways. You know what, I can't read a physical map anymore.
I wouldn't even know which direction to look at it. Why? Because I've been using GPS so long that to get from point A to point B, if my GPS is not working, I'm kind of stuck.
The longer we use GPS, the less likely it is that map reading will be one of our skills. There are organs we don't use as human beings because we don't need them anymore. There are skills we don't use as human beings because we don't use them anymore.
If you don't use them, enough, you will lose it. I'm worried we're going to lose our capacity to reason if we don't stop looking up the answer to everything. So reason it out because that's something again that we have to work on as human beings.
Which brings me to my fourth point something that is going to sound strange you've heard the old saying that an empty mind is a devil's workshop right probably written by a puritan who wanted everybody to be thinking about something all the time but let's face it an empty mind is also where the most creative most amazing things happen i call these aha moments where as human beings you connect things that are unconnected and come up with marvelous insights i don't know what our comedians was was thinking about when he slipped into his bathtub thousands of years ago. But he jumped out of the bathtub, ran down the street naked, and hence was born the Archimedes principle. I know what Galileo was thinking about when he sat under the apple tree, but when the apple fell on his head, he could have cursed the apple like most of us would have. Gravity was born. So in a sense some amazing things have happened from human beings taking something that completely unrelated and making that leap.
I am not at the same level as our Archimedes and Galileo, and my insights are not as, you know, not even close to being as deep, but I'll give you two examples of things that have happened because I've let my mind wander. The first was more than a decade ago, living in New Jersey still and we had a blizzard and there was snow piles feet high outside the house and if you've ever lived in the suburbs and there's snow outside your house and on the road you got to shovel the thing you're required to so I go out there with the shovel it's hours of shoveling and while I'm shoveling all my neighbors are also shoveling you know but their kids are all out including mine are all out in the snow playing making snowmen throwing snowballs at each other laughing. So here we are, the adults cursing with our snow shovels and kids laughing.
And as I was watching how snow evokes these very different reactions, the same snow, very different reactions in two sets of human beings, I thought of the analogy in investing. When things happen and the same thing happening can create misery for some and joy on the part of others. In fact, I used it as a base. basis for a post on value investing versus growth investing, where I said value investors are like the adults with shovels and growth investors are like the kids with snowmen.
I'm not sure who's having, you know, who comes out ahead in this analogy. But it's a post that I wrote called Snowmen and Shovels, which remains among my favorite posts because it connected something I was doing, shoveling snow to something completely unrelated. The second post, more recently, happened a few months ago.
ago. Now that morning I woke up to an email from a reader of my blog in Iceland and he'd emailed me about a valuation challenge he was facing. He said I'm valuing a company called the Blue Lagoon. It's an Icelandic spa and it's in the pathway.
of a volcanic eruption. Lava is flowing towards it and I'm not sure how to bring in the risk of that volcanic lava flow into my valuation. There's nothing in the books about a volcanic lava risk adjustment factor. I didn't have the answer but my dog started pawing at the door because he needed to go out.
So I decided to take the dog for a walk. I didn't take my phone with me, thank god. And while I was walking my dog I started thinking about the lava flowing towards Blue Lagoon.
And while I was thinking about it, I thought about another news story I'd read a few days ago about this company called 23andMe, which collects genetics from saliva, cultures from people, and then comes up with a genetic makeup of the person. But they did a data leak of every genetic sample they had in their computer. Catastrophic. And finally, I started thinking about fossil fuel companies. especially in the face of climate change and how investors were pricing in or not pricing in the possibility.
Three very unrelated stories but as I was thinking about these stories because I had nothing to do I started thinking about how human beings collectively deal with catastrophes. Are we rational? Do we bring in the probability of catastrophes and the consequences of our decisions and while I was thinking about it I said look I own a house that that's two blocks from the ocean for which I paid a lot of money. And it's on one of the biggest earthquake falls in California.
Did I bring in the possibility of an earthquake or a tsunami into how much I paid for the house? And the answer was absolutely not. So it's not something you can point a finger at other people saying you're being irrational.
Human beings, when faced with catastrophe, tend to ignore it. And they're not being irrational, and here's why. If there is a true...
catastrophe are you really going to be checking your portfolio in fact if you ever watch the Mad Max movies and I use this example in my catastrophe post is there anybody in that movie checks what their portfolio is doing Not really, right? You're surviving. The truth is, in the event of a catastrophe, what your portfolio does becomes irrelevant.
So it becomes actually rational for us to keep that out of the picture when we do our analysis. My more general point is I was able to do all of this because I was walking my dog and I didn't have a phone and nothing to do. You know what, I look at each of those four friends and I'm going to argue that it's getting more difficult rather than less to practice those four things.
What are the four things? Being a generalist. connecting your weak side or working on your weak side. Third, connecting the stories because you have an empty mind, because you have nothing to do.
On each of those dimensions, it's getting more and more. and more difficult to do what we need to do. In fact, if you're a conspiracy theorist, you could argue that what technology companies in particular have done over the last few decades is taken away our agency by doing what? Again, you've got to be a conspiracy theorist to believe it.
They've essentially, who created Google Search, so your reasoning powers would weaken, who created the smartphones, you'd have no time to have an empty mind. In many ways, what's happening is becoming more specialized, more one-dimensional, using our reasoning powers less, and we have less and less free time. I may be delusional on my part, but I want to keep my bot away from me, which means I'm going to be a dabbler, which is what I tend to do.
I write about what interests me. I'm not a specialist. I'm open about admitting to it. I'm going to work on my weak side. I'm a number cruncher.
I'm going to work on my storytelling. I'm going to try reasoning my way down. answers before looking them up online.
I want to take my dog for more walks. I mean the dog's not happy about it. He likes to sleep. But I think that those are all things that I need to do to keep my bot behind me. So as you think about this process.
I know that I'm in a fairly unique place because everything I do is in the public domain. There's a bot with my name that's coming after me. For me, the AI thread is here.
I can't deny its existence. I can't push it into the background. You may not be there.
Your work may not be public. There may be no bot with your name on it. But I think you have to act like there is a bot with your name on it coming after you. And if you think about strategies, there are three possible strategies you can try.
Remember, one reason I'm exposed is everything I do is in the public domain. I've chosen to keep it in the public domain because I wanted to. You're saying, you're stupid.
You shouldn't have kept it in the public domain. Maybe if I had been more secretive about my writing and my thoughts, it would be more difficult for the demodern bot to match me up. I have no desire for that secrecy, but maybe your job lends itself better to that secrecy.
Maybe you can do your job behind a cloak where bots can't see what you do. As you think about this strategy where you keep your work secret enough that an AI can't watch what you're doing, there are two reasons you have to be wary. First is, there are other people doing what you do.
So if you're an equity research analyst, you can be secretive about what you do. But guess what? AI can still watch other equity research analysts do what they do, and it's going to figure out what you do. The second is, even if you're secretive about process, your actions give you away. Let me give you a quick example.
give you an example. If you're an active investor, you might have a secret sauce that you use to pick stocks. But if I can observe what you buy, when you buy, what you sell, and when you sell, I can reverse engineer what your process is. And AI is smart enough to do it. So try the secrecy strategy.
Maybe it'll work for you. It won't work for me. The second is what I call the system protection strategy. Let me explain. I've bought and sold houses multiple times in the US in the last few decades and every time I've done it it strikes me how much waste there is in the process how many intermediaries there are lawyers realtors title deed checkers the list goes on every one of them gets paid and I keep wondering why do we need them And the answer is very simple.
It's not logical. There might have been an original reason we needed them, but the truth is, with technology, we don't need them anymore. But the system requires that they be there. In other words, your house ownership will not transfer if you don't have all of those intermediaries in there. The system protects their jobs.
At System Protection for Incumbents, it's not just in real estate, it's everywhere. I mean, where is it written? that only universities can grant degrees and certificates. I'll tell you where it's written.
It's written in regulations that if you try to give a degree without getting certified, the state is going to crack down on you. It's great system protection. So here's the second protection you can get from AI. Get the system to protect you.
In the appraisal business, as I said, it's not that AI cannot do what appraisers do, but appraisers will put pressure on courts and accounting rule writers. make sure that only appraisals where human appraisers sign off will be accepted and AI appraisals will not. So second, go for system protection.
So try to be secretive, go for system protection. I think they're both risky strategies. Here's my advice for whatever it's worth.
You know how in business we talk about moats. Moats are competitive advantages and make it more difficult for other people to do what you do. Companies with bigger and more sustainable moats are worth more than companies without those moats.
Why should that apply just for businesses? Think about moats at a personal level. There's you and your bot and the question you're asking is what can I do that my bot cannot do? do? What is your sustainable advantage?
You're thinking, I hope. I would love to tell you what the answer is. But I can't tell you because it's your job and you have to come up with the modes.
And I'd be lying if I told you I knew what my modes were. But what you're seeing in this post is my attempt to start laying that out. And just as a bottom line, I have to be quite honest, I am a work in progress.
My bot is a work in progress and I want to be ahead of my bot. It's my human nature. I'm going to try to create a niche. that is mine.
And this may be an impossible mission but I will keep you posted on how it's going. But here's I think the disconcerting part. It's entirely possible that my bot will get much better at what I do than I am in which case it might be writing the next blog post and you will not know the difference. And while I work at fighting my bot, here's what I'd like you to do. I'd like you to act like there's a bot with your name coming after what you do.
And I'd like you to think about what you can do to stay ahead of your bot. And the answer is not complicated. Rediscover what made you a human being.
Rediscover humanity because therein lies the answer to beating your bot. I hope you found this session useful and I thank you very much for listening.