UT CATT Panel

Brad also hosts the Behind the Ticker podcast, featuring interviews with ETF issuers and service providers. He holds a bachelor's with honors from Duquesne University and completed his coursework at NYU for certified financial planner designation. Additionally, he holds Series 66 and Series 3 licenses. Brad, welcome.

I have Sudhir Hala from My Stock DNA and Tal Swartz from AI Funds. So, gentlemen, thank you very much for being here today. And why don't we go down the line, maybe starting with Kyle, give a bit about your background and the company that you're here and the company that, you're working for. Yeah. Thanks, Brad. Thank you, Kumar. It's good to be with everybody. So, my name is Kyle Wiggs. I am the co-founder and CEO at a company called UX Wealth Partners. Just a brief bit of background, been almost 20 years in the wealth management industry, spending a lot of time with financial advisors who work with the community, taking their feedback, and ultimately running, broker-dealer platforms from technology to investments to sales and distribution.

And it was along that journey that I realized our industry was not keeping up with other industries in terms of bringing things like artificial intelligence to the clients that we ultimately serve. And so, we launched UX Wealth Partners as really a dedicated platform, bringing trading, billing, reporting, to be able to support and service some of the strategies that you're going to hear about today. Because a lot of the legacy technology providers that exist just really don't have the capability to deliver some of the modern strategies that you're going to hear about today. Thanks, Kyle. Maybe we can go over to Sudhir. Hi, everyone.

And the reason I believe that is because, and this is before ChatGPT and everything else, is because I saw what AI is able to do with narrow AI in many, many different domains. And the fact that it was able to exceed human ability in games and other areas. So the idea was, why can't it not do that also in finance? So that's the reason we started AI funds with the idea that this is really the biggest trend that's going to happen over the next few decades. And it's an opportunity that we all need to take advantage of. Great.

What I see somewhat working effectively today. The second part would be where I think we're making a bit of a mistake as an industry. And I think to Tal's comments, what's the opportunity? Where are we missing it today? And what does that future look like? I think in terms of where we sit today, what's my observation of what's effective is you think about like a client resource management system. Presumably most people, in today's session are familiar with like a sales force. We're seeing the utilization of artificial intelligence to gather and organize data and information that I think is effective. I think it can always get better in terms of how these various technologies interact and potentially integrate with one another.

But that, to me, is effective. You think about things like gathering large volumes of data and creating real-time insights. you're walking into a client meeting. Let's say you're preparing for tomorrow and you've got an overwhelming, volume of data and information. And I think going through that, organizing and pulling out real-time insights certainly works and is effective today. And then the last one is, as Tal talked about, things like chat GPT. We're starting to see some of the advisor community really, I think, improve their communication capabilities. And so I think that is working well today. It can always obviously get better. One of the huge mistakes that we are observing is none of us like to get into a service call where we get into an AI loop.

That has created a lot of frustration. There are some firms that are in the industry today that are scoring really poorly in terms of customer service. And I think there is still that human component that's really important, that some of the ways AI is being utilized today is ineffective. What's the opportunity and where are we headed? It's the investment piece. And we're going to talk a lot about that today. But when I travel the country and I sit down with your traditional chief investment officer that's running a multibillion-dollar firm and they have an investment committee, they are still using old ways to try to solve new problems. They're sitting down, they're pouring through information and data, and they're using humans to make investment decisions.

So Sudhir, why don't you go first? How do you see AI redefining the investment landscape and kind of where are we going in the future?

And then you use that to drive decision-making. Efficiency, this is, again, straightforward, right? So, just like computers help you be more efficient, AI is going to help you be even more efficient. You'll replace the capabilities of an entry-level analyst. And that's where we see a lot of the focus of AI, especially right now in finance, in terms of improving the efficiency. The third aspect, which is managing emotions. Now, like it or not, for most of the retail investors, which is what our community deals with, the financial advisors, financial markets are a lot about emotions. It's about the emotion of fear and greed, right? So fear when the market is going down, what the hell is happening, what's going to happen to my profile, portfolio, and so on and so forth.

And greed when somebody else is making money or your neighbor is making money, and then you have the case of envy.

So that's equivalent of managing risk. Now, for us, within MyStockDNA, what we are leveraging AI for is primarily the first dimension, which is taking into account large volumes of data to account for whether we should, whether it's a small ditch or a big ditch and how should we avoid the ditch. And the second is being able to unemotionally take that decision. So the emotion part of it is a significant component in what we have built. Great.

If you look at the large cap investors, human investors, something like 85% of them underperform the S&P 500. And this is published data by S&P Standard Course. So we know that humans tend to underperform the benchmark. Okay. And the reason has to do, like Sudhir said, has a lot to do with emotions. Okay. Because unfortunately, we were not designed to play the stock market. We're designed to, the revolution to survive in the savannah. And when we see a lion, we know we need to run away. So when the market starts falling, guess what? We want to run away. And that ends up hurting our performance as human managers.

And as Sudhir mentioned, AI does not have emotions. So it's actually able to take advantage of those situations and actually generate alpha during those situations. So I think the biggest opportunity for AI today is really on the active side, because there there's already underperformance and people are still, there's more active investing than passive. Although passive has grown a lot over the past few years. But there's a huge opportunity there. Because AI can outperform humans, potentially outperform human managers. And I think that's really the direction the market is going to go first. Simply because AI, for a bunch of reasons. One, it can process a lot more data than humans can. It makes unemotional decisions.

And it's able to do a better job predicting both potential return and potential risk. And as a result, you're able to build smarter portfolios that will do better both in up markets and down markets. So that's a high level view of where things are going, at least in my view. And we can dive deeper into how specifically we think we can help make that happen.

There's a bunch of these indicators. They go back through time and they find what is the most similar time to today. And based on that, they make a prediction about what's going to happen now. So they usually take one or two or three data points from the past and try to do that by using visual analysis or something of that nature. And what we've done with Bela is actually try to do that on a systematic basis so that we don't just take one or two times. We take every similar time over the last 50 years. So in a sense, you're getting a distribution. Now, if you take the most similar fingerprint of the macro economy today and you match it to the past, you end up figuring out that, the match is not just 1979.

It's also 1973 and 1968 and so on. So there's a lot of times in the past. And sometimes the market went up and sometimes it went down. So you get a distribution and then you can make a prediction from that. So the macro side can actually tell you over the next few weeks, and that's the horizon that we're making a prediction. Over the next few weeks, one to four weeks usually, is the market more likely to go higher or go lower? Okay. Once you know that, once the prediction is made, and this is a statistical prediction, it's not 100% right, but it's trying to make a prediction about bigger drawdowns. So drawdowns of 5% or more. So, if it's a small blip down, that happens all the time.

It's the bigger ones you're trying to avoid, right? So it's making a prediction on that. And then the next stage is now we have a portfolio optimizer that figures, given what we know where the market is going, then what is the right portfolio? Okay. What is the right mix of assets, given the client's risk level? So we match it to different risk levels. And then we have a particular portfolio that's designed for that. Okay. So it's designed with the right mix of assets. And, we have the traditional 60-40 portfolio for, an investor that may be in the middle range of risk aversion. Well, you tell the AI, well, given that level of volatility, what is the right mix?

And sometimes it does have 40% in bonds, but many times it doesn't. It has other assets that offset the correlations and reduce risk. And we can talk more about what is risk really for investors. It's really the downside risk. So you have to also optimize for that in a smarter way. So there's a lot of things we can do as, the AI can do to help us build portfolios that are smarter. So you have those two parts that Baylor does, the macro strategies, the portfolio manager. And at the end, you get a better overall portfolio performance out of sample, not just in backtesting, but also in real-time performance.

So we're able to avoid that and then come back in and capture the upside. So that was this year. Every year is different. I can give you an example. In 2022, it went risk off many, many more times. It was a downward year and Baylor was, I think, out of the market much more than was in the market. Okay. So it really depends on the year and the level of volatility and, and the signals that it's seeing in the different indicators that it's looking at. But typically, it's every week. And in general, there's a rebalance once every four weeks because the patterns in the underlying assets have changed. And it's typically an opportunity to improve the portfolio performance by rebalancing every four weeks.

So these are active strategies. They don't necessarily trade every week, but they trade about once a month typically. And then they go usually risk off twice a year on average, I would say, and an average year. But it changes year to year. This year was only once so far, and we've been bullish the entire year. Great.

And instead of the weather conditions, you have market conditions to maneuver. But the challenge remains the same. You need to get to your destination not just fast, but also safe. So Darwin, as an AI, does something similar. What it does is it takes into account all of these market conditions, those different lanes that you're navigating in, and increases and tries to maximize the return to risk, which means you want you to get to your destination faster, but also safer. Now, if it's possible in the technology world to kind of take the leap of faith of self-driving cars, then the same analogy holds true for your portfolio, if you construct it similar to a construct of a video game.

And that's what AI does. For us, it navigates the different asset classes and lanes and then tries to get to its destination and outperform the index in the process of doing so.

Yeah. So one of the great joys, but also challenges, is taking the complexity of what Tal and Sudhir and others have built. And how do you get that into the actual client account? Because to get from there to there is actually a really complex problem to try to solve. Maybe an analogy for the group to think about. There are some brilliant architects in the world, and sometimes they design things, but then the general contractor has the task or the role of building it. And sometimes the concept as an architect is wonderful, but it's not always something that can be implemented. And I think trying to eliminate as many of those sort of friction or pain points as possible is what UX is solving for in the marketplace.

How do we get what these guys just described as closely as humanly possible into the end client account?

Great. So let's move on to segment two. And I'm going to open the floor up to all three of you. So anybody can jump in here and I'd like to keep this a little bit more conversational, which would be, look, we just learned how AI can create really sophisticated strategies. we touched on some of the main benefits, but really, how is AI solving some of the many pain points for investors that traditional investment paradigms just can't solve for? Anybody is welcome to take that one first. I can jump in.

Turns out it's portfolio and something called the efficient frontier. And there we're taught. I used to teach this to MBA students. We teach that as an investor, you need to get to the efficient frontier. You need to get to the highest point, meaning the point with the highest return and the lowest risk. And that's volatility. Well, guess what? That was an amazing innovation. But since then, we've learned a lot. We've learned that humans don't really try to optimize volatility. In fact, what you're trying to do is minimize downside risk. So the downside volatility. And they don't mind upside volatility. In fact, they like it. So the more volatility is on the upside, the better. So really, what we're looking for is a type of an asymmetric return.

And as it turns out, that's extremely difficult to do by hand. In fact, you need machine learning and artificial intelligence to solve these type of models. So once you figure that out, you realize, really to match what we as human investors are looking to optimize for, which is this asymmetric return where we're trying to maximize the upside volatility and minimize the downside volatility or the downside risk or the drawdown or whatever way you like to think about that risk. That's really what we want to do. And the only way to really do it is with computers, because it's extremely complicated, especially when you have a universe of many assets and you're trying to build a portfolio. And when you look at the math of it, where you have every asset and you have hundreds of assets and every asset can have a different weight,

Turns out the universe of potential portfolios is huge. It's more than the number of sands of grain and planet Earth. So it's a huge number and you can only really do it with a computer and machine learning can solve for those portfolios. It's very hard to do, but it's possible. So that's kind of, I think, an area. And that's why this paradigm shift needs to happen, because too many portfolio managers are still running portfolios on min-various optimization. They might look at the Sertini ratio, which looks at the downside risk. But typically, they don't really know how to optimize for that or for much more sophisticated ratios of risk and return. So I think that's the direction we can go. Sudhir, to you.

I wanted to completely agree with you. And I think we have Professor Kumar there. He's also going to agree, right? So see, the Markowitz efficient frontier made sense at a point in time, right?

And in the process, they kind of destroy their own return. Now, is it possible using AI to scale portfolios to such a level where you cater to individualized emotional risk frontier? So the level at which, let's say, Brad, you will panic will be very different from the level at which, say, a Kyle will panic or a Tal will panic or a retail investor will panic. Can I create portfolios managed to that return risk profile that we talked about in such a way that it caters to your individual emotional risk frontiers? And I think that's something, for me at least, that becomes very interesting.

You think about this meeting today. This meeting in the current construct doesn't happen when Markowitz rolled out with his theory. And so we've seen so much sort of evolution, if you will. I think it's time that our industry really embraces it. We did a paper at a prior stop for me probably about 10 years ago. It was done by one of the portfolio managers at a company I was at. And the title of the paper was Volatility is Not Risk. And the analogy that he drew was he went out and he surveyed thousands of investors. And he gave them basically two options.

But like, what are the challenges in designing your particular, mousetraps, if I could, that can help it respond to unprecedented market events and, really ugly volatility. So either of you happy to take that first.

My portfolio performs great. And that's really the biggest challenge as we go on. So whether you call it back testing, curve fitting, forward bias, server bias, you can use a whole lot of terms. But these terms are just as have implications for AI as much as it is for humans. Right. So as humans as well, we tend to make biases based on what we have learned in the past. So one thing that I want to emphasize here is that it's not that you can eliminate forward bias or survivor bias or the curve fitting. You can reduce the impact of that and reducing the impact of that has a combination of two things. AI can be used to predict. AI can be used to react. Everyone in the in the last two days have talked about predicting, predicting the earnings, predicting the next stock, predicting how the market will do in the future.

So reacting is as important and reacting using AI is more important is as important as well. So I don't know if you guys have heard that there's this this experiment which was done called the Turtle Traders. So it was Richard Dennis as a trader who basically trained several novice guys out of the street. And within a couple of weeks, he trained them to become to manage their portfolio and they were pretty successful. So so for me, it is about not just addressing the prediction part of it, but being able to address the the forward bias, server bias, survivor bias and so on and so forth in the reaction part of it. So what we've done on the my stock DNA side is we have two monkeys.

We have the blindfolded chimp, which essentially picks stocks at random. And then so the theory is that if a model can perform well with random stocks, you can be reasonably certain that it performs well in the future if you're throwing random events at it. And the second is a is a Netflix inspired chaos monkey. So what we do in this is we we essentially go and randomly change returns from positive to negative returns. And that's, again, to test the efficiency of the model in terms of managing the downside risk. Right. So as long as you balance these two, the prediction versus reaction and you're able to do it, I think that that addresses the point that we were talking about. Tal, anything to add there?

Yeah, I do. So, with Bela, what we do essentially is we train it over time to learn what are the patterns that indicate that the market is more likely to go down than up. So in a sense, it's doing what a classical machine learning, a paradigm of classification. And this is something that's very it's a classical thing. But the way to solve it for the market, it turns out not to be classical at all because it's a very chaotic system. So you need something that's extremely robust. And in some ways, because it's robust, that's what makes it more successful, because if it does find a pattern, it's more likely to be repeated.

Now, it's still statistical. So what that means is, yeah, it worked in 2008. It predicted that it predicted 2018 and 2001. And in between, it also sometimes predicts things that don't happen. And sometimes it predicts things. It misses some drawdowns that do happen. But if it can capture most of them using this evolving and learning paradigm, then it's more likely to work in the future. Now, there's no guarantee, obviously. But you can kind of you build confidence over time because you can see that the patterns it detects tend to repeat. They don't always repeat the same, too, because the factors change over time.

Because what's important today wasn't important even a year or two years ago. So, inflation now is relatively high. But it's lower than it was a year ago. But it's still higher than it was two years ago. So things change. And what's important to the model also evolves over time. So I think that's important because it's not exactly it's an evolving system that's learning over time. That's what the L in Bela. And the idea is that it's trying to find those patterns that tend to repeat. But at the same time, they have to be extremely robust because you don't want things to change too quickly over time. Because these things, they need to work now. And they also needed to work, five and 10 years and 15 years ago.

So you want something like that to work for you. And really, it's a statistical game at the end of the day. It's not going to get everything right. But it's going to get enough of the drawdowns right that's going to make a big difference for your portfolio. That's the expectation. And that's what we train it to do. If you don't mind, I want to add on to that. So I did a study of all the famous analysts that we have heard of, right? so take the list. And I did a study on terms of how accurate were their predictions for the following year. So at the end of 2019 for 2020, at the end of 2020 for 2021, and so on and so forth.

And we did this with Merrill Lynch, Bank of America, and so on and so forth. And these are well-known firms who are making a significant amount of money in the market, right? So the test that we were trying to do is whether the efficacy of their prediction has an impact on their results. Turns out that there were seven analysts out of, I think, 15 that we studied who got it right in 2020. There were seven analysts who got it right in 2021. There were four analysts who got it right in 2022. But there were zero analysts who got it right both in 2021 and 2022. The problem is that most of them had their inherent bias. The analysts who were bullish were bullish throughout. Analysts who were bearish tended to be bearish throughout.

So half glass full, glass empty. you continue to be that. So if these guys are making money, it's not just because they're predicting something right or wrong. It's also they're able to react when their prediction goes wrong. And the ability to, so what Pal talked about, right? You can look at past patterns and you can predict. But what is also important is saying, okay, you recognize the fact that, okay, what I've predicted hasn't turned out true. So how do you react? How do you learn? And how do you morph?

The answer is it depends. there's no doubt that some of the large institutions are now utilizing AI to, I think, organize unstructured data in a way that they can make intelligent decisions. That is happening, and it's happening at scale for them.

That myself, you, as well as Tal is involved in, is you're using MyStockDNA and Darwin to take blends of different AI managers with different thoughts and blending them together in a portfolio matching someone's risk. So in about two minutes, could you kind of explain that project and how it can benefit the end retail or end client?

Well. And it happens, it'll happen with Tal, it'll happen with me, it'll happen with anyone else. Unless we evolve, it's not all of our strategies are going to survive in the future, right? And we don't want to be emotionally fed to, emotionally linked to those strategies. Right now for an advisor, when he goes in, he puts his trust on a model. He says, I want to pick manager A or manager B or manager C simply because they have a reputation, they've done, spent a lot on marketing and so on and so forth. What if we change that paradigm to say that Tal, me and the other managers out there, which is including you, Thor and the other AI managers, we compete for the right to manage money in

A portfolio. If we do well, we get more money to manage. If we don't do well, we get less money to manage. And what if this competition is governed by AI as opposed to a bureaucratic investment committee, which is basically taking into account a lot of other emotional variables in terms of who they're helping, who they're not hurting in terms of taking the decision. What if we emotionally take the decision as to which manager do we allocate funds to and how much? And that's the logic with which we're coming up hopefully soon along with UX Wealth, an all-star team consisting of all of these managers who are essentially all these AI managers who are essentially competing with each other for a position in the team.

And then once they're in the team, the idea is that they collaborate with each other to increase the alpha and reduce the beta for the portfolio.

So where I would love to see AI make a big headway is in terms of democratizing that 401k process, because by virtue of doing that, you are going to make an impact on a generation, which is going to be able to retire early simply because you managed funds well. Beating the S&P 500, according to me, is not a problem. Anyone with a little bit of common sense and a knowledge, I can train them to do that. But doing that effectively by managing your emotions is a big thing. And then getting past all of this, what do you call it, this marketing terminology, right? When you get a brokerage statement, you get like 100 lines of saying, okay, this, that. Getting past that and using AI to do that would be my dream project.

Great. Tao? Yeah, I think Sudhir nailed it. It's really, if you're talking about in the short run, and I'm saying the next five to 10 years is short run, AI can already, in my opinion, do better than active human investors. So I think there's a huge opportunity there. And because these are active strategy, generally they generate, they could potentially generate short-term capital gains. So an ideal place for them are in tax advantage accounts. And that's why 401ks and IRAs are a really good place for these strategies. But unfortunately today, most investors can't access them from those types of accounts. So I think that kind of shift will happen. And those strategies will grow in those retirement accounts.

And I think what you'll see over time, at least my vision, is that AI will penetrate and will take more and more assets also for the passive investors. And this will happen over the next few decades. Finance is a very conservative industry. So I think it's going to take time until assets move over. But it will happen. And you'll see that more and more also in areas like ETFs and others. So I think that's really the path that we're on. And we're really just at the beginning. Because also the AI we have today is still, you have to say it. And other people in AI say this, this is the dumbest it will ever be. So they're still, they're pretty dumb right now.

And it's really the narrow AI that's very, very smart today. And that the general AI is still very dumb. So it's still going to take a while until you see general AI that can replace a human. I think that's not going to happen for a while. And I think the human is still in the loop for quite a long time, for decades to come. So I think it's a combination of human plus AI that we're going to have here for a long time. But you will see AI managing assets in a much more, a bigger way over the next few years. And it will grow to billions and trillions over time. Kyle, any last thoughts or are you ready to move to questions?

That, to me, is the big idea that we have to try to figure out how to solve. Okay. Well, we've got 28 questions. I don't think we have time to get to all of them. So go ahead and give us your best questions. Yes.

Because there's a saying, events that have never happened in the past always happen, right? So are we hedging for that at all, or are we completely dismissing out and saying, okay, if it happens, we're going to be surprised, so be it. So I can tell you from my experience, we were running Bela in 2001 during COVID, and we reduced risk during that period. So Bela noticed that something was happening. I don't know what the pattern is exactly that it noticed, but back, I think it was in early February, it actually noticed of 2001, it noticed that something was going on, and it reduced risk. And it wasn't 100% risk off, but it was low risk right before the market corrected.

And it went back in a few weeks later. Same thing happened in 2022, where there was a decline. It wasn't risk off the whole year, but it was risk off for major parts of the year. So now you're saying, COVID never happened before. But guess what? A lot of the patterns in financial markets were similar before COVID happened, or at least waves were happening. You can call it ripples in the financial data that AI could detect and say, you know what? Something here is happening. Of course, you couldn't tell me it was COVID. It could tell me that there's a higher risk of a 5% or higher drawdown during this period. So I want to reduce risk exposure.

So that's something it can detect. It can detect those patterns. It's not necessarily that the risk is the same or the black swan event is the same. But many times, the markets are already showing, the markets are some of the best predictions. So, and it takes a while for investors to switch mindsets, but some investors could tell something was happening with COVID. So they were doing some things in the markets. They were creating some patterns. So you can see those patterns. So at least the AI can see it. I'm not sure a human could see it, but the AI could see it and could actually react accordingly. So I think that's my answer why it could potentially detect these black swan events. Now, will it detect the next one?

I don't know. There's no guarantee. It's only a probabilistic system, right? I would say the probability is positive that it would, but no guarantee. So for me, the answer would be slightly different. So imagine the Matrix movie where there are bullets flying around. And I use this analogy often, right? So there are two ways of avoiding the bullet. One is you have complete confidence in your prediction ability to dodge the bullets, get in the room and dodge. And the second is don't get in the room itself, right? Or you can also do the thing of getting in the room and getting shot by a bullet and then deciding this is not it. this is not safe. And I want to stay out of the room.

Now, it's the same with AI. As long as your AI is not wedded to its prediction and is able to manage the fact that its prediction can and will be wrong, and you have the reaction ability to be able to react, you're safe. Now, can it dodge a black swan event like the Yellowstone erupting or something like that? Probably not. Not. But can it dodge a black swan event like the market crashing 10%, 20% in the next one month? Yes, it can. So it's a probability thing. Okay. So moving on, maybe a couple of questions. I'll wise out. So managing money, as you've kind of touched upon, is a two-sided job. One is how does the money get invested in the market?

And when a human actually manages money, there's this other big part of managing your clients. And so we've touched upon this. The client's emotions, particularly greed versus fear kind of emotions, handling them. Now, we've talked a lot about how AI is going to manage money in the investment science side. Have you all thought about or are aware of any work on how AI can be used to manage clients, their emotions, reporting, that kind of stuff? Kumar, the bald patch here, that comes from managing emotions. But on a serious note. So emotions are individual specific, right? So your emotions and how you react to the market is going to be very different from how somebody else reacts.

So my hypothesis here is that it would be impossible for AI to manage your emotions, but it would be extremely possible for AI to build something that caters to your emotion, right? So I cannot influence your emotion. I can educate you. I can educate an investor. But it's going to be – but that education can only go so much distance. So I think a large part of what AI can do in the future is also construct a portfolio or a product which caters to how you emotionally react to things. And that's where I think it can help.

But at the end of the day, you have to trust that the AI is making the right decision at the right time. And here's the statistical performance so far and talk about all those numbers. And I think when you talk about it that way, you take the emotion out. They realize, there's nothing really I need to do here because guess what? The AI is making a better decision than I or my advisor could probably make because it's unemotional and it's based on statistics. And at the end of the day, that's the best that we can do with investing. So I think that's a stronger case than the advisor saying, look, I just made this decision. I know it's losing money, but trust me because we know that humans underperform in these situations.

So I think that's the better discussion. One last question. And this is an interesting general note. So we've talked about AI replacing lower level skill sets in the investing industry. And we do understand that to be able to get to a higher level of skill sets and expertise, we do need humans to go through this lower level to graduate and, get to the higher level. So if you're talking about a world where AI is going to replace all the lower level, skill acquiring or skill learning positions, will we ever be able to actually go, have humans gain the experience to get to the top? It's a good question. So I think, look, we're still in the early stages of AI and the influence on society.

And I think we're seeing some of the effects today. So today, AI is not replacing coders. It's making coders more efficient. So if you're a developer, you can develop much, much better code. But it's not yet developing software on its own. if you're looking in some areas, yes, AI is giving us a boost in terms of our abilities, things that we can do. And it's natural that in finance, that's an area where it's, it's all statistics and numbers. That's an area where AI and machine learning should do better than humans because we're not the best in figuring out numbers and doing all that stuff in our head, especially when there's billions of data points to work with. So that's an area where I would expect AI to do really well.

And I think there's some areas like that where AI is going to augment us and make us better, just like your phone makes you better, right? so it's another piece of technology that improves our lives. That's the way I like to think about it. And I think human passion for different areas, it's going to drive really what we go and learn and study. I'm passionate about portfolio optimization and AI. So that's an area that I excel in. And I think each person here has their own level of passion of what they care about most. That's going to be the motivation, the real fundamental motivation of what drives us at the end of the day.