Trading Tomorrow - Navigating Trends in Capital Markets

From ChatGPT to Market Manipulation: The AI Revolution in Capital Markets

• Numerix • Season 4 • Episode 1

Share episode

Artificial intelligence has evolved from a cutting-edge concept to a vital force in capital markets—reshaping strategies, streamlining decisions, and making prompt engineering a must-have skill for today’s financial professionals.

In this episode of Trading Tomorrow - Navigating Trends in Capital Markets, we revisit key AI trends and bold predictions from our first two seasons. What did we get right? What surprised us? Tune in as we reflect on AI's rapid rise and its real-world impact on the financial landscape.

Speaker 1:

Welcome to Trading Tomorrow Navigating Trends in Capital Markets, the podcast where we deep dive into technologies reshaping the world of capital markets. I'm your host, jim Jockle, a veteran of the finance industry with a passion for the complexities of financial technologies and market trends. In each episode, we'll explore the cutting-edge trends, tools and strategies driving today's financial landscapes and paving the way for the future. With the finance industry at a pivotal point, influenced by groundbreaking innovations, it's more crucial than ever to understand how these technological advancements interact with market dynamics. Hi, I'm Jim Jockle, host of Trading Tomorrow navigating trends in capital markets.

Speaker 2:

And I'm Emily Drew, the podcast's producer.

Speaker 1:

This is our fourth season of Trading Tomorrow. Over the years, we've had fascinating conversations with people from all over the globe who are leading the way in new technologies and trends within capital markets. Now, for the next few episodes, we'd like to take a step back pause and really reflect on some of the conversations we've had over the first and second seasons of our show, alongside the current state of the technologies we've highlighted over the years.

Speaker 2:

And we figured the best way to kick things off was with artificial intelligence, one of the most powerful forces redefining capital markets here in 2025. What's especially interesting is that we launched this podcast back in 2023, which is really right after ChatGPT burst into the public consciousness. In many ways, we've had a front row seat to the rapid evolution of this technology as it's reshaped the industry and our lives in turn.

Speaker 1:

Well, you know, from high frequency trading to AI driven risk analysis, machine learning has moved from being a competitive edge to an absolute necessity. But, to quote Spider-Man, with great power comes great responsibility. And what happens when AI begins to make decisions faster than human traders can? How do we balance automation with regulation, and what's next for AI and finance? And we'll discuss all of that today. Ai has proven to be a major focus over the past few years. In fact, it was the topic of our very first episode back in September 2023.

Speaker 2:

Wow, jim, that feels like a lifetime ago, and yet it's amazing to look back and realize how foundational that conversation was.

Speaker 1:

It really was, and in the evolution of AI it was a lifetime ago. But during that episode we heard from Parag Sharma, global Head of Artificial Intelligence at Citi's AI Center of Excellence, and for me there was a few major takeaways from that interview. Let's roll that tape. How do you keep up with the speed of growth, adaptation and innovation in this space?

Speaker 3:

Simply put, we do not keep up with it. There's too much going on to keep up with everything, but the idea here is not to know about everything, but to focus on the aspects of artificial intelligence or generative AI, which is a subcomponent of that. To focus on the things that matter to you specifically for your job or the industry that you work in, and really spend some quality time getting under the hood, understanding what the technology does, but also understanding the use cases. So a number of ways to keep up to speed with depending where you are in your life cycle, in your career life cycles.

Speaker 1:

Well, you know, I, for me, Emily, I think one of the key things was Parag was absolutely right there is absolutely no way to stay on top of the evolution that's coming out. You know, we've always made that old joke it's, you know, evolution, not revolution. But in the world of dealing with AI, it is a revolution every major release. But you know, I think one of the fascinating things was you know, while Parag was running this center of excellence, Citi as a whole at that point in time was still very restrictive of the use of AI for all of its employees. And now City, in many ways, is a leader among financial institutions in terms of embracing AI across the entirety of the organization something that felt so niche back then has become really second nature at this point.

Speaker 2:

I love this moment because it really captures how quickly we've adopted and integrated AI into our daily workflows. As you said, they weren't allowing employees at Citi to use AI in their daily lives, but now it has just overtaken so so many things that we do, and it's become so important to understand how to talk to these systems effectively.

Speaker 3:

Firstly, we need to all have prompt engineering in our job description. The sooner the better. So anybody listening in, please do a course online and get that prompt engineering going. Joking apart, the reason actually you need prompt engineering is like I was explaining. You have this room-sized model, quote-unquote, and you're interacting with it. The clearer you are in your interaction, the better results you're going to get, and prompt engineering is simply that. How clearly can you articulate your ask so that you get the right?

Speaker 1:

response. Well, at this point I think I should be able to give up my CMO title and just put prompt engineer. But I think there's a couple things that have really evolved. Number one yes, absolutely, people need to be able to have an understanding of prompting and how to get the most out of AI, and there's still so many people that still think of it as just a search engine on steroids.

Speaker 1:

But the other thing that I really think is fascinating is the innovation of tools that have come out since 2023 that are really, let's call it, persona-based, in the sense of tailoring AI interactions to utilize multiple large language models, but really begin to tailor to a particular job, function or role. I can even say, you know, as part of the marketing team, we've utilized certain tools that are built for marketers, fit for purpose. The AI knows how to interact and get the most out of what we're trying to accomplish with the tool. So, while prompt engineering still is something that everyone and I stress everyone should be embracing, and you don't even have to take online courses at this point, there's so much information out there and YouTube videos and whatnot for free. But the way we've seen that evolution evolve over two and a half years is to make AI even more approachable where that skill isn't necessary. But I urge, if you're going to get the most out of these tools, prompt engineer is something you need to be.

Speaker 2:

And it's so interesting because I you know, back when we first had this conversation with Prague, I think that prompt engineering was something we all were nervous about. It seemed really scary. You felt like you had to take classes, like you were saying Jim. But now you know, my mom knows how to talk to ChatGPT. She looks it up on YouTube and on TikTok and both her and my dad are comfortable using ChatGPT and I think it shows how adaptable it's become, that really everyone is able to use it now, and I remember at the time, prompt engineering it felt kind of like a new language, but now again it's practically a core skill set, whether you're in research, compliance, trading and knowing how to frame your question and get actionable insights.

Speaker 2:

It's critical and we've seen how a slightly better prompt can drastically improve the output you get from a model. I mean, I personally have seen it and I've found my footing in that model. I mean I personally have seen it and I've found my footing in that too. I mean it's a lot of trial and a lot of error, but at this point I think it's crucial. You need to be trying, you need to be failing, you need to be looking these things up and you need to get comfortable because AI is not going anywhere.

Speaker 1:

You know, Emily, we only spoke about AI three other times during season one.

Speaker 2:

That's surprising because I felt like it was everywhere, even outside of work. I think it just speaks to how quickly the conversation was accelerating behind the scenes.

Speaker 1:

It really did, but we did speak about it again fairly early in the series during our episode with Neil Chennai, operating partner at Sandhill East. He covered four tech trends you need to be tracking and AI was one of them, and I think he even picked AI for his trend drop. We now made it to the final question in this podcast and we call it the trend drop. It's like a desert island question. So if you could only track one future technology within capital markets over the next few years, what would it be and why?

Speaker 4:

Well, I kind of think it's a combination of the advanced cloud computing being the cloud and generative AI and how that's going to change how people do their job, and a lot of it's to help drive productivity. Research I mean, if you look across the whole research in the medicine field, in the pharma field, it just seems like that over the next three years let's pick that number that generative AI is going to provide really, really great advances in research.

Speaker 1:

I don't know if I would go as far to say, as Neil's comments were so prescient, but definitely had his finger on the pulse of where the market was going and I think even when we had that conversation it was still early days. My own personal exploration was fairly low, but now the things you're hearing in medicine and logistics, supply chain and finance as we talk about all the time, I mean it's staggering the advancements that have been made in just three short years.

Speaker 2:

I think even just in the later episodes of this podcast that are going to be coming out, we've heard some things that AI, especially angenic AI, is able to do. That is again, like you said, Jim staggering. I mean just stuff we couldn't have even dreamed of back when we were recording these very first episodes, back in 2023. And what I appreciated about that conversation is how Neil framed AI not just as a trend, but as a shift in how people work. You know, it wasn't about replacing people, it was about expanding human capability, and I think he made a great point about how AI would accelerate productivity across industries like medicine and research, like you mentioned, which we're absolutely seeing now, and I mean for me, in communications, my output has almost doubled, While I still, you know, need to read through and approve everything that goes out those rough first drafts, they can be handled by AI, which cuts my production time.

Speaker 1:

Which makes me happy, exactly so. We've also had another conversation on AI and ML and on investing with Chauvin Jane of Okon, but we're going to hold on to that because she's joining us again and it's going to be exciting to speak with her again on a one-on-one basis, but also just for her perspectives on how things have changed.

Speaker 2:

So that brings us to our conversation with Broadridge.

Speaker 1:

Yes, and so, as you may recall, they sent out a digital transformation study in 2023. Although digital transformation does seem slightly overused, that's what it was called, but they really took a hard look into the cutting-edge technologies captivating the finance industry's investments and attention, and you might remember, one of their findings was that 71% of their respondents said AI is changing the way they work, and here's what they had to say about it. And here's what they had to say about it. If you had a crystal ball, how do you think this number has either grown or shrunk since that?

Speaker 5:

I would say it's probably increased. I think I'd feel pretty confident saying that it's increased. You know to your point. You're using it every day. I'm not yet. I work in a strategy type role, so could I see it potentially replacing the creation of strategic plans and things down the road? Yeah, so maybe I'm just avoiding accepting future? I don't know.

Speaker 2:

It's interesting because, again, we really were at the forefront of Gen AI when we started this podcast, and episodes like this remind me of that. I mean, broadridge's more recent 2024 study showed that firms plan to increase AI investments by 21%, which is the highest across all categories, even beyond blockchain and robotic process automation. The previous year's study had cloud and cybersecurity beating. Ai, and robotic process automation was tied. I think it really shows growth. Companies have seen the importance of AI and they're adjusting accordingly, and Broadridge also developed their own internal AI tool, built on large language models, which they rolled out to employees. They casually mentioned it in our interview and it's great to see that it's taken off internally. I mean, it really shows how companies are not just exploring AI, they're embedding it.

Speaker 1:

Well, I would even say in conversations with friends and individuals on the street. You know, even those who have been laggards are now in the throes of adoption. And you know, I recently read a McKinsey study that I thought was fascinating, whereby it was saying AI adoption is really driven by middle managers and individual contributors, and it's actually been executive management that has been a laggard right. If you go back to 2023, when we were having this conversation, it was something along the lines of 80% 90% of all AI and machine learning projects fail which was curtailing management's investment into these projects.

Speaker 1:

But now what we're seeing is this overarching trend where people want to use these tools they're differentiators for them in the ways they work which is now creating this nexus where management is saying, hmm, I understand, we're getting these productivity claims internally. I have external forces, whether it be investors, whether, you know, depending on what your ownership is private equity stockholders. They want to know what your AI initiatives are as companies. So you're seeing these two things kind of come together at this point where, if organizations are not embracing these tools, you're going to have unhappy employees. Right, they want the benefits, they want to explore, they want to utilize these contributions different place. I think Broadridge, in terms of exploring the way they used AI internally and embedding it into their own ways of working and processes and in their products, is kind of a way forward that many companies are at today.

Speaker 2:

And I think you hit the nail on the head there, jim. When we started trading tomorrow, back in 2023, ai was something that it was a buzzword, it was exciting, it was in all the news, it was fun, it was interesting, but it wasn't tangible yet. And now, in just two short years and really a little less than that, because this was filmed near the end of 2023, it's now embedded, it's important, it's crucial and it's a part of pretty much every single company in some way or another, and AI continued to be a hot topic throughout season two, which took place in the beginning of 2024, which was right after season one.

Speaker 1:

Well, at that point we really dove deeper into AI and we were remiss if we didn't. You know not that other emerging technologies are not as exciting, but AI has been obviously the most captivating. You know, so many of those conversations explored why AI rose in popularity so rapidly and how significant that impact has already been. From Kevin McPartland of Coalition Greenwich to Adam Hyland, a PhD student at the University of Washington, we got some incredibly rich perspectives. Let's roll that tape.

Speaker 6:

But if we were to compare right now the impact, the potential impact, say, in the next five years, of blockchain, when it sort of first came on the scene, versus the conversation about AI now, I'm a lot more excited about AI, and part of that, perhaps, is because you can see it and feel it. Once there was a consumer application that allowed you to do these incredible things and have a machine answer you and give you results, write code for you, write press releases for you, whatever that really, I think, opened a lot of people's eyes as to what this could do. Now, ai obviously isn't new. I think I took an intro to artificial intelligence class in college in the 90s, which I don't remember any of. But, all kidding aside, the immediate impacts feel much more gigantic.

Speaker 7:

Yeah, the adoption and, like you said, I think integration is a really important. Distinguishing feature of how much this sort of blew up on the scene is that not only were people adopting it, but people were linking it into other processes, like connecting the chat GPT to a text-to-speech generator or or the reverse right, connecting it to to voice to text and then sending that to chat GPT. Like all of these multimodal components coming together is something that we have not seen at this speed before, just ever.

Speaker 1:

I think it's fascinating now in terms of Adam's thoughts on multimodal Today. That is the advancement where we're seeing some of the most gains, whether it be incorporating video speech in addition to text photos, if you haven't tried this, and now if you have Apple Intelligence, it's integrated with Siri. Open up the refrigerator, take a picture and ask ChatGPT three things that you should make for dinner. It's absolutely fascinating. So I think it's getting to the point where multimodal is commonplace and it's really changing the way we're interacting with AI tools, which is really really became changeable and just going off of your refrigerator comment.

Speaker 2:

I recently wanted to redecorate my apartment. I took a photo in an AI app and I was able to say what needs to go where, what do we need to buy, and tell me the background behind why I need to do this. So give me a reason, Give me a decorator who's talked about this and it was able to do that, which is incredible and really it's just. These systems are evolving so fast and I think it's something that we've seen in real life too recently through the development of advanced multimodal AI systems like Alib. It's compact design. It can potentially run on smartphones, which enhances device capabilities and user interactions. And one of my favorite conversations from season two was with FinPilot CEO Lakshay Shuan. Described as chat GPT for financial questions, FinPilot was using AI to extract insights from unstructured financial data and at the time that felt so groundbreaking, but now we're seeing even more companies trying to build tools just like that.

Speaker 1:

And this kind of brings us back to that prompt engineering statement. I think one of the things that was really interesting when we saw what Finpilot was doing was a lot of the prompting behind the scenes. So it allowed the end users to really interface without being a prompt engineer to be able to just ask a simple question. But behind the scenes these very, very elegant prompts were being designed because it had the context of that persona, I think. One question I would like to go back to see how FinPilot is doing, because, as these LMMs are getting stronger and stronger, they have a lot of these capabilities. Am it's learning from my interactions? Even though I'm not training models, because you know I don't want my stuff out there in the world, you know I can give it the personas that I want and want to interact with. So I think we've had some early adopters that could have challenges as general purpose tools get even stronger as general purpose tools.

Speaker 2:

Get even stronger. Have you done that challenge yet that people are suggesting you do on social media, where you go and you ask ChatGPT what it knows about you?

Speaker 1:

I have and it's the whole kind of FBI, cia kind of thing. It is a little scary but you know, for those who are just passive ChatGPT enthusiasts, some of the Reddit groups are fantastic, and you know they, those who are just passive chat GPT enthusiasts some of the Reddit groups are fantastic, and you know they come up with the craziest things.

Speaker 2:

And speaking of what chat GPT does and doesn't know, perhaps one of the most important conversations we had during season two was with Professor Michael Wellman of the University of Michigan.

Speaker 1:

Yeah, the regulatory conversation. That was an important one. Let's roll the tape on that.

Speaker 8:

But there's also a potential for new AI loopholes.

Speaker 8:

The current laws and most regulations are built under the assumption that it's human beings making the decisions, and when it's computers making the decisions, it could be that there's something about the way the regulations are written that lets them get around it. And one example is in the area of market manipulation, where a lot of the existing rules promulgated by the SEC and through Dodd-Frank in the United States rely on determining the intent of a trader when they put in orders to the system. Are they intended to really trade or are they just there to mislead? Well, there could be some question about how you judge intent when it comes to a computer program, and, in particular, if your computer program was generated by machine learning, or that is, if your trading strategy was generated by computer, through machine learning. That might seem to provide one some kind of deniability as to intent. Now, that's a loophole, I think, and actually one of the pieces of legislation that Senator Warner and colleagues just filed, the Financial AI Risk Reduction Act, does attempt to close that particular loophole.

Speaker 1:

Basically you're suggesting the AI independently can learn potentially to manipulate markets on its own.

Speaker 8:

That's right. We've actually demonstrated that in our own research and I wouldn't say it's a huge surprise. But we basically set up a system where we told a trader that we wanted to maximize its own profits and made a certain environment where they had an interest in a benchmark in some derivatives. That depended on some pricing benchmark and it learned to manipulate the primary market to move the benchmark such that they made additional profit on the benchmark, even though we didn't tell it to do that specifically. It basically learned to sacrifice profit in the primary market in order to make more on the derivatives and that would be considered a kind of manipulation that was not directly programmed into the system but was rather learned automatically.

Speaker 1:

You know, obviously, regulation is going to be a guiding principle and this is really a white space because a lot of regulators don't even know what they're dealingils are. And having reward systems within these LMMs to ensure that ethics, and having guardrails inside of these LMMs that are very transparent to ensure that the activities and outputs are ethical, I think is the foundational statement.

Speaker 1:

I mean, I'm sure if you listen to other podcasts there's plenty of commentary that AI is going to rule the world and we're you know, years away from Terminator 4, or it might be closer than we like to think, but you know, clearly the regulatory environment is something that we all have to watch, especially now that we're in a new administration. Obviously, ai is on the forefront. You know, obviously, another there's been push to bring manufacturing and power back into the United States. There's been push to bring manufacturing and power back into the United States not to go into all the tariffs and the politics of it, but AI is playing hands across the global community.

Speaker 2:

Yeah, that clip still gives me the chills. I mean, it was really the first time I truly could start to comprehend how AI could learn market manipulation, and not because it was told to, but because the incentive structure allowed it to. I mean, it's a good reminder of why thoughtful regulation and ethical design are so critical in this space, and it's a scary reminder that it's something that we really need to continue to push and look into, because I think one of the few places we've seen a lot of evolution in the past two years since we started filming this show is in regulation, which is considering the conversations we've had over the past half hour about how much chat, gpt and AI and how these things have all changed so much in the past two years, I feel like regulation hasn't, and so I really think that that's somewhere we need to look moving into 2025.

Speaker 1:

Yeah, and especially as we look ahead to this next season and some of the conversations we've already had, agentic AI is now on the forefront and if you think of, it's disturbing that market manipulations could happen and whatnot what happens when you let the AI work autonomously. So that's something we're going to dive into this season because that is the rage of the year.

Speaker 2:

Agentic AI is one of the most fascinating things we've spoken about during this season. I can't wait until people get to listen to those episodes. I mean, it is absolutely where all the new innovation is going to be happening, and I think people don't really understand that it's happening yet. So even I thought this was a conversation that we're looking at technology five, six, seven years out, but you know, we're looking at technology that's happening right now.

Speaker 1:

Well, emily. Thank you, it's been a fun ride so far. I can't wait to have more great conversations and, as we move forward in season four, please make sure you continue to tune into the show as we explore AI and other trends and technologies that are changing and shaping the capital markets.

Speaker 2:

Thank you, Jim.

Speaker 1:

Thanks so much for listening to today's episode and if you're enjoying Trading Tomorrow, navigating trends and capital markets, be sure to like, subscribe and share, and we'll see you next time.