 Hi, and welcome to Bright Minds from Tickmill. I'm your host, Patrick Munderling, and in this series, we're setting out to answer some of the most commonly asked questions around investment and trading through entertaining and insightful conversations with seasoned insiders. Back in episode six of this season of Bright Minds, in our discussion with Hariton Christou, we learned the current state of play with AI in the financial industry. And we heard how Tickmill specifically are using AI to provide better tools for retail investors. In today's episode, we want to build on that discussion and hear what the future holds for the role of AI in trading and investing. Artificial intelligence research is developing at an incredible pace, thanks to an ever-increasing computational power, growing availability of data, and near constant innovation, which is expanding the areas in which algorithms and machine learning can be applied. As more and more industries begin to see the transformational potential of AI and machine learning, demand is increasing and investment into AI is skyrocketing thanks to the ability to address complex challenges. The numbers and patterns-based data-heavy nature of the financial industry aligns perfectly with the problem-solving strengths of artificial intelligence. Therefore, we should expect to see huge revolutionary disruption in every area of the sector, from sentiment analysis and risk management to automated trading. As a consequence of these changes, we may also see a drastic impact on governance, regulation and ethical considerations within the space. So where will the biggest and fastest changes occur? How can investors prepare for the future? And with all the hype around this new technology, will it be justified? We'll be speaking about all these questions and more with today's guest, Tal Schwartz. Tal is a serial entrepreneur, quantitative researcher, technology innovator and former professor of finance. Over the past 20 years, Tal has founded numerous companies including Clicktail and Expand Beyond and his current focus, AI funds, aims to provide best-in-class AI managed investment solutions. Tal, thanks for joining us today. Could you start things off by telling us a bit more about your career so far? As an entrepreneur, I actually started my career wanting to become a scientist and wanting to work with lots of data and using algorithms. I wanted to become an astrophysicist and then fell in love with finance as an undergrad while being a part of a student investment club. And one thing led to another, got a PhD in finance, worked in hedge funds and taught finance and it's been kind of something I love and been involved with over the past 25 plus years. So it's kind of all led to what I'm doing today with AI funds because using a lot of data and machine learning to figure out ways to improve on just the standard passive investments that most of us do. Broadly speaking, what are some of the key ways in which AI is transforming investment strategies and portfolio management right now and how is that impacting individual investors? There's different levels that AI has been used and actually it's been used for many years already in finance. It's just that we're not aware of it. Specifically, it's used in a very high frequency trading and things that take fractions of a second in the way orders are routed and the way different entities trade those orders. So that uses a lot of algorithms and machine learning because simply it's too fast for humans to react. Where AI has not been used as much is in areas where most humans invest in, that's time scales of minutes, hours, days, weeks and months. In those scales, humans and passive strategies still dominate because AI has not simply not had the power to be able to affect that until now. So that's all been changing recently and I think that's where we're going to see AI involved a lot more in asset management meaning you can have AI run strategies which is what we do with AI funds is really have the AI which we call BALA. It stands for Bayesian AI learning algorithm. So BALA, our system, that's what it does. It manages investments over those time scales. And I've noticed you've written online about the hype cycle for emerging tech and how that applies to AI specifically. First, I guess what is the hype cycle and where are we right now within AI's development and what does the future hold? Yeah, so this is a concept that was started by Gardner which is one of these consulting companies that provides a lot of information for many Fortune 500 companies. So they started the concept of a hype cycle where the idea is that technologies start growing slowly and then as they accelerate they go through this massive ramp up and a peak of hype of what the potential is and after you hit the peak eventually there's a drawdown where it comes to the through and then comes back up and that's when you actually get the value from the technology. So if you can map many, many technologies the internet, the dot-com boom, all of those things have gone through this type of hype cycle. AI is actually relatively the beginning of that because we're just now starting to understand what the potential is for what it can do but it's going to disrupt many different industries in many different ways, financial and other areas in our lives and we're just at the beginning of that hype cycle. So I don't think we're still there. Usually you get to, you know you're near the top where infalluation is going through the roof and they're like we had in the dot-com and we're not there yet. We're far from that and I think we're also early on in the development of the technology. We're still in the early stages. What we just saw with chat GPT is just the beginning of what we're going to see. So there's a lot more to come from, you know, really the early stages. How important is it to be able to understand and interpret the actions of an algorithm that's running AI? It's something that is extremely valuable. Unfortunately what happens is as these models get bigger and bigger and they get more and more complex it's very difficult to do that. If you think about chat GPT it has like a trillion different parameters and when it gives you an answer it doesn't come from one place. It's distributed across a very large network and it's hard to isolate and figure out exactly what's going on there and this is true for other AI models as well. That's why a lot of times they're called black boxes because it's hard to know really what's going on. Obviously it's preferable. There are models which are much more clear and it's obvious what the relationships are between the inputs and outputs and where how things are affected but usually it doesn't much simpler. So they're unable to do the complex things that we really want AI to do and so when we want AI to do something complex it becomes very complex and almost too complex to really understand to the details. Thinking about that complexity I guess what's your view on the ethical implications of using AI in trading for example and what ethical considerations should be taken into account to to I guess really ensure fair and transparent use of AI algorithms or is it just going to be a race to an endpoint where whoever can throw the most money at the problem is going to have the best solution hence the best algorithm hence the best returns. I would say okay let's take step back and think about historical perspective. Technology keeps improving our lives and every time we have a technological improvement that has a big impact on our society. So think about I don't know the automobile that came around there were horses and carriages that had a big impact people who were riding horses and people who made all the equipment for that they lost their their their livelihood and you had new industries pop up that improved quality of life for humanity and you keep seeing that over and over as new technologies get get developed and this is no difference up until now recently most of the improvement has been in physical strength meaning you make we developed machines that could do things you know faster better than us this is like the first time that we actually have a technology that can actually potentially think better than us or process information in a way that's more better than us and help us improve so it's it's kind of a paradigm shift and that's why everybody keeps talking about how this is going to be such a big impact on our on the world and the reason it's a big impact it's it's going to be it's a huge economic improvement to potentially to what we have today so let me give you an example okay not in finest but in a in real world like a call center a call center today is filled with thousands of people and what do they do they get a question someone calls in and they go through a list of answers and they answer the question based on that list AI could do that AI could do that much better than a human in fact they can process all those thousands of potential answers and then answer it in a much more efficient way so that a call center is actually going to be much better than the best human run call center that we have today all those people working there are going to be out of a job unfortunately okay there'll be other jobs for them but this this specific job that's this kind of mundane repeating of information that's going to be gone maybe initially it's only going to be the tier one support and we already have those but they're not very good today when you call a call you kind of annoyed when they answer the phone right imagine not being annoyed but being so pleasantly surprised that it answers all the questions and it's so friendly nice that you actually rather speak to a robot than speak to a human because there's so much better and answering and knowing all the information that's the future we're getting towards so it's it's going to be a paradigm shift so there are ethical questions because people are going to be losing jobs first off okay that's part of this transition and this transition unlike other ones as we've seen with Moore's law things are accelerating exponentially so transitions are now going faster which meaning our society is going to struggle with that transition because it's going to happen much faster than we used to so if things took years to take and we could absorb the layoffs and all this stuff is going to potentially happen much faster which means many more people are going to suffer in the process but the driver here is really the economics of it i mean it's just the potential economic impact on all our lives is huge i mean we're looking at a multiple x improvement in quality of life for humanity in general so that's the direction this is going now specifically in finance yes there are ethical questions because if you're a human trader and now you're trading against the machine you're at a disadvantage the machine knows a lot more that has a lot more data sources and it can react much more quickly than you can so it's going to have an advantage on you and we already see this today because there are quite a lot of other hedge funds and other entities that use trading bots and things of that nature that give them an edge and we're going to see that over time that edge is going to improve over time and this is why I started AI funds I believed in in the future AI will manage the vast majority of investable assets i think that's the future we're headed towards because this is exactly the kind of area where AI will be superior to humans because this huge amount of data you need to make objective decisions based on past probabilistic distributions of what has happened and AI can do that better than humans can and they can do it quicker and without the emotional baggage that we have without the fear and greed that we have it's something that's going to happen and the reason i'm very sure of that is because when you look at areas where there's games adversarial games like chess poker go you name it AI today is the world champion there's actually no it's better than the best human and the best humans and in playing those games actually learning from AI had to play it better so you have AI teaching us now how to play better the game and that's going to happen also in finance AI is going to teach us how to invest better because it can see so much more than we can and just on that point about more investable assets in the future being controlled by AI what does a market that is increasingly run by impenetrable super efficient AI actually look like I mean look over time look the AI is going to have an edge over humans okay the systems in the short term the the the fractional of a second those already run by algorithms and those become more smarter over time as as as the AI gets better at processing macroeconomic data and other information then it's going to start trading longer term holding periods you know hours days weeks that's what we do at AI you know at AI funds we actually take a bunch of macroeconomic data right we compare what the current macroeconomic fingerprint looks like compared to the last 50 plus years and then figure out like what the potential is for returns in the future now AI is going to get better and better at doing that and it keeps learning based on different time scales and more and more data at the end it's going to dominate any human trader it's going to it's a mess and eventually you'll find that AI is trading AI meaning that the market is clearing via algorithms humans are just going to be not as quick and not as mad I have as much information as the AI now you're saying well up until now you know AI might have only been accessing data on prices but now it can access also written data and spoken data and all the data that basically the vast amounts of data that exist in the world that humans had access to and AI did not so over time all this vast amount of data is going to come in and it's going to influence these algorithms in a same way to influence us humans as new information happens as new news happens it will be able to react faster and more accurately than human can I guess one of the one of the issues with respect to human interaction within the markets that we've seen over time of the numerous bubbles and busts and flash crashes do you think that AI and machine learning will be useful in the future in terms of being able to detect and mitigate risks such as flash crashes and market manipulation yeah so I think it's really interesting the thing about bubbles is that we can it's a very human thing you know investment bubbles we can actually replicate those in a lab and in financial and behavioral labs we can recreate bubbles which is fascinating so it's something that's almost in us built in and I think one of the reasons AI is going to outperform humans is because it can take advantage of all these cognitive biases that we have built in so it can actually just like the youtube algorithm can cause us to stay more and click through more and watch more video by feeding the things that we want to see the same way in trading the algorithm will figure out how to cause a bubble and then take advantage of it while it's driving all the humans in to bid the price up while it's selling so it will it will actually learn how to play us because we can be played we're human right so it will know how to create fake momentum and fake crashes and take advantage of all those things to take to make sure that it can maximize its profits for its owners for whoever's running that AI so I would not be surprised if you'll see exacerbated fluctuations in the market because it benefits someone who knows how to take advantage of it and it draws people in oh look how high this is going and how fast and let's buy because it's going up and so we have these these biases in us and and AI will take advantage of it unfortunately and that will drive eventually human traders out because they will realize that they're you know they're losing it's like a casino think of it like a casino the AI has a 0.1% edge over enough time the law of large numbers they will they will you know they will win and we will lose so did you tell us a bit more about your company AI funds and and I guess what you're aiming to achieve with it essentially what we try to do with AI funds is simulate what a human analyst would do and what a typical human analyst would say is okay this is what the picture looks like now in terms of the economy and what the markets are and this is most similar to 1970 whatever six and back then the market went up and or it went down or did this and they make some conclusion based on this one point in time so we try to do that across all periods and all points in time so we say let's take a lot of macroeconomic variables find create this macroeconomic fingerprint then go back through time figure out all the times that were most similar to today not just one time all the times and then aggregate all those times together create a distribution of what returns look like when you run time forward and then based on the shape of distribution the AI figures out well is it more is the market more slightly more likely to go up he's more likely to go down he's more likely to go sideways and then based on that prediction of where the market is most likely to go and again this is probabilistic because when you go back through time through all those past times sometimes the market goes up sometimes it goes down but you have a sort of probability you can say okay 60 percent of the time goes up so it's more likely to go higher based on these probabilities it constructs portfolios that there are designed to do well in that type of environment so it's a dynamics strategy active strategy which changes and shifts in the market based on what it's learning and based on what it's seeing so it could be bullish and then if it sees signals that tell it you know what there's too much risk right now it can reduce the risk level and adjust dynamically adjust to the risk that it perceives in the market in the potential upside what's for you over the the next few years is going to be there the single biggest transformation that you'll think we'll we'll witness in the financial industry the low-hanging fruit of things we're going to see now are things where let's say mundane activities are going to be automated with AI so simple things that have to do with portfolio rebalancing and things that have to do with customer interaction like we said call centers and things like that simple things are going to get automated first that's going to be just the world becoming more and more automated more and more simple anything that a human has a repetitive task that's going to get automated and typically is going to get replaced with AI where I think the big value is going to be is in actual asset management today that's the vast majority of that is done by humans and it's done by portfolio managers and there's a lot of interaction with customers around that I think over time that's the shift that's going to happen as AI takes over a lot of those types of roles and you'll still have humans running a lot of these things but most of the analysis is going to be done by AI is going to come out with the insights of why we're taking the strategy we're doing and why we're doing it this way and there's going to be the more long-term portfolios that are run by AI where it's kind of relatively static and there's more active strategies like the one I'm running where it's actually trying to figure out when to reduce or increase risk based on the signaling the signals that are coming from the market so there'll be different types of strategies run by AI and that's going to revolutionize I think asset management so if today you think about asset management being having two pillars one passive where essentially you just invest in usually market weighted you know passive indices and you have active human strategies where humans are deciding what to trade and how to do it if you think of those as the two main pillars there's going to be a third pillar emerging which is the AI run strategies and that's going to eventually be larger than the other two it's going to take years and many years for this to happen but that's that's the future we're headed towards and I guess in terms of bringing this back to the tick mill platform and how retail investors can engage with AI because obviously your standard retail investor one may not have the the investment funds significant enough to access a fund that is running AI strategies or they wouldn't have the ability certainly wouldn't have the ability to to create AI strategies independently is there a model whereby smaller retail investors can engage with the type of strategy you're running either via subscription or some type of alert service or is it purely that money needs to be put on deposit with your firm and then it's run automatically so actually we offer our service as a subscription to investment advisors so investment advisors are actually getting our signals for the different strategies right and then retail investors are able to invest through those RIAs so it's really available to retail investors but you have to go through your advisor and work with them directly and I think that's the the the right way to do it and it's the right way to actually access millions of potential investors that can benefit from this. Tal, thank you so much for your time today and for joining us it's a it's certainly a fascinating and very very topical issue at the moment where can our listeners find you online Tal to follow up with with any questions or to follow your work indeed? Sure absolutely so they can join connect to me on LinkedIn and follow my my posts there and if if they're interested in contacting me directly they can email me at tal at aifunds.com. Excellent. I look forward to hearing from them. Thanks again for your time today Tal. Thank you, thanks for having me.