The Intelligence Economy What Happens When AI Gets Real

[Music] Good morning gentlemen and uh welcome to the stage. Uh let's get right into it. Um David, you are a computer science PhD from MIT and in your company to sigma that you have founded, you use a lot of um high power computing um machine learning, distributed u uh computing to do a lot of your trading strategies. Can you help us um uh do a deep dive and say which aspects of all these computing is considered AI? That's a uh in a way a funny question because today uh the important thing to understand is all computing is AI because if you say that you're using AI everyone feels that you're leading edge your valuations go up and so forth.

So AI has become essentially a buzzword for computing. But that really is uh in a way a a uh just a funny way of saying that AI is just a whole bunch of different techniques and approaches that people have actually been using for 30 years. Today AI is largely to this group I would imagine large language models chat GPT that's something that's actually quite different from uh other kinds of machine learning approaches. AI is a really big diverse field but it's become in a way uh narrowed down by uh uh the the current um uh cycle of companies like OpenAI and and Anthropic deploying in large scale large language models.

Right. Well, hopefully we'll be able to unpack um what you have just said later on. Um, Bill, your company that you founded, SSNC, you've got a solution called AI Gateway that helps your customers um, integrate their work processes into AI into their work process. Can you elaborate on how that works? Well, one of the things that we think is really important is if you're going to start using AI that, you know, in a big way that you need to have controls and you need to be able to keep the AI agents that you have doing a lot of things for you.

It has to know that there's boundaries. So, our AR gateway has guard rails where we don't let it out of certain parameters that we have set up prior to allowing it to really solve a number of problems for us and speed up our processing and allow it to grab data from all over the place. But but always in a controlled manner that that our people can feel good and our customers can feel good that we're doing it in a controlled manner.

Right. I want to sort of you two gentlemen to take a look at the uh products and the work that you have. Do you use computing just to come up with a range of probabilistic solutions for the human to make a decision on or does the computer actually take over the thinking for you? Well, I think uh in in in the uh use cases that I've been involved with at my company, it's always been a partnership between the humans, the people, and the algorithms. Uh and and that's I think an important point because to effectively use computing to effectively use AI, you really need to have the the right people that can form this sort of partnership.

AI technologies are not turnkey. You can't just deploy it and have it magically transform your business. you can't magically use it to generate new ideas and new profits for a company. So the the partnership is what's really important. When I founded uh my firm uh you know this partnership between technologists and business people uh and the algorithms was a key uh you know the key concept really uh in in building the whole organization and over the years the partnership between man and and machine or person and machine has only grown stronger.

Right. William, is that the same kind of um experience with your clients?

I I think it's a little different because, you know, David at two sigma is constantly looking for returns and looking for ways in which to take advantage of opportunities around the world. you know, we we basically are making sure that all the trading and transaction processing is reconciled and delivered back to our clients in ways they want. So, the things that we use AI the most is we might get a credit agreement and we might get 2,000 credit agreements in our clients and we'll have AI agents that will scrub those credit agreements, give us all the relevant information and allow us to move that through our systems and get it back to our clients a lot faster than we could if you did it manually. We also do it with you know general partners like David really want you to have the limited partner statements perfect right like don't spell your name wrong Eugene or right

right right you know that's really unpopular so we'll have an AI agent that scrubs all the limited partnership statements checks all the names checks all the addresses checks all the numbers and that's really effective for us

but to David's point earlier on where everything is AI, but what what you just just described is really just high powered computing.

Well, it can think, right? If it doesn't have something, it'll go to another document and find out, you know, did you change your address? Did you put in a change of address form? You know, did you have a new child?

So, it can reason.

It can Yeah, it can go in it. It and it it can do it fast,

right? you know which is you know if you have hundreds of LPS you know it's something where that can take a lot of time if it's done manually which we did it manually for a long time

okay I want to quickly touch on the point of risk uh weighing of risks theoretically you can calculate the probability of risk down to a trillionth of a of a percent right uh but at the end of the day it is still up to a lot of humans to make an emotional um uh gut instinct where does gut instinct or emotions factor in this world of uh AI and probability right now?

Ah you know in a way you're asking a very philosophical question. Uh so I've thought quite a bit about what is um a gut instinct and in the end I can't really uh uh figure it out. So, I think that people that have a good gut instinct usually when they say, "Oh, well, I think this is what's going to happen." They're they have to be right more often than they're wrong. Right?

That's what it means to have a gut a good gut instinct. And I'm not sure that it they're being random. I mean, we don't really understand how how we think. It may just be that they're exposed to a lot of data. this is a human and they integrate the data in their brain and an idea pops into their head and then I'm going to do that.

That may not be just a gut instinct. That may be the result of some very thoughtful thinking that's going on deep inside your brain. So now you look at uh an AI system. You've all used chat GBT or uh for example sometimes it will kind of give you the impression that it has a gut instinct. And why would that be?

Well, because it's taking lots and lots of data. It's kind of integrating it together using uh large language model technology and then it it just says this is what I think. Okay, that may not be so different than human gut instinct. We don't know. But the short answer is if you think that the results from a large language model are are are good.

If you think that it's gut instinct is usually right on certain topics and so forth, it doesn't really make a difference. Is this human gut instinct or AI gut instinct? Probably not a big difference in my opinion.

What do you think, Bill? Well, I think David's point is is that, you know, we all talk about intuition and things like that. And, you know, for people that aren't in finance or don't manage portfolios or don't manage their own money even, you know, sometimes they talk about their returns, you know, and and someone will tell you, "Oh, I have 50% returns." I says, "Yeah, me too. When I go to Vegas and put it on red and it happens to come up red, right?"

But one might think that the next time I go to Vegas, my return might be not 50, it might be minus 50. Right? So, you know, again, as you look at patterns, you know, as you see recognition, I mean, at SSNC, we've done 84 acquisitions and so you start seeing patterns, right? And you start understanding that, hey, this makes sense and this doesn't make sense. And I don't know that there's anything other than you know sort of intuition or gut instinct similar to what David said before about you know the the large language models and AI can start looking at those acquisitions and piling up hey these are good these are not so good and pretty soon you start getting a pattern that says you know Eureka let's do this right

right and so it's both humans and compute power

working together. Okay, one more um philosophical question before we get down to the topic at hand which is what happens when AI gets real. Let me ask you this. Do you think um AI will become self-aware or achieve sentience as they call it?

Uh you know uh probably if it does we'll never know. uh you know it's an interesting you can just think about the person sitting next to you in in the audience you know are they really sentient or not um you don't you you you you're going to just expect that they are but you have no way to really prove it and so this I think is a pretty deep philosophical question where uh you know if you were to ask you know an LLM are you sentient it could very well say yeah I am uh and uh you You know, you might say impossible. A machine can't be sentient, but you know, I think in the end it's going to be a question that doesn't have an answer.

Sounds like a bit of a copout. Well, look, I uh the one thing I'll say, I've always believed that the human mind um is essentially a computer, a biological computer, and that whatever the human mind is doing, eventually it will be replicated in in some kind of digital computer. And and so if we replicate the human mind completely, then I guess you end up with a sentient machine.

Yeah, I think the same thing. I I think you know you can go back 30 40 years and there's 2001 a space odyssey and h how seemed to get pretty smart there for a while and start spouting orders at the humans you know and it it kind of gave everyone a charring effect and and same thing today you know you go back say 50 years or 60 years 70 years and they have somebody have an ailment and they bleed you. So that that seems to be a tad um barbaric in 20 25 terms. Whereas today, you know, if you get a cancer, you may get burned, right, with with radiation or, you know, maybe you get poisoned with chemo, but something tells me 25 years from now, they're going to look back at this and say, "Boy, that was barbaric." Right?

So as learning happens and again all these things that the computers are constantly layering on I think pretty likely it will be sent and it will think

okay let me ask you this. So, how will AI change um the world that we live in? Right? Back in uh the late '9s before uh the internet became ubiquitous, nobody can imagine life as we know it today. How the internet has upended our changed our lives, changed society, changed the way we work, the way we consume, the way we play, etc.

, etc. So given what we know about AI now, looking forward 10 years, 15 years, 20 years from now, what will the world be? Would you care to speculate?

It it it I think it's a lot harder to imagine than people might think. Um, you know, none of us were around, but if you go back to when electricity uh was was being uh commercialized, um no one would have imagined uh something like this behind me. And the the electricity obviously has transformed everything in ways that were unimaginable. People were thinking about light bulbs, you know, oh, electricity, we don't need candles anymore. Then you think about the internet, as you said, in uh uh you know, the mid '9s.

Uh no one really imagined that it was going to become like electricity. It was going to be ubiquitous. every there's no way we can turn the internet off today any more than we can't turn off electricity because we depend upon it for absolutely everything. It's an important fabric of of our lives. I believe that in in 10 15 years you know probably not not not five uh AI is going to be like that.

It will be ubiquitous. We just depend upon it for everything that we do and it will be impossible to live without it. And so, you know, you're not going to really think about it as a standalone technology. You're just going to think it's just like electricity. We need it.

Yeah. I I I believe that that's 100% true. And, you know, today, if you think about what's really changing your life or your children's life or others, you know, it's it's it's probably your cell phone. You know, when when I was growing up, it you really didn't want to lose your wallet. Now, you really don't want to lose your phone, you know, and those kinds of things kind of change people's lives.

I mean, I I went to school from Indiana to the University of Hawaii when I was 18, 6,000 miles from home, and no telephone.

Let's zoom in. How will it change the way we work? How will it change the way we work? We know that from a lot there's been a lot of discussion that menial repetitive work will will be will be redundant, right? So it will free us up to do more creative work.

Is that what will happen?

I think it's a little bit different. Um I think and again this goes back to imagining something that has not yet come to be. So the nature of work is just going to change quite a bit over 20 years. And and and so jobs are going to look different. We're going to work in different ways.

Um, it would be too narrow to think that AI is just going to take the existing kind of jobs that we all have and make them more efficient. That that's not what the internet did. It created a whole new category of jobs. It reshuffled the economy. Just look back at, you know, from the 90s to where we are today.

Everything is completely reshuffled. And so AI is going to do the same thing. Many people say, "Oh, the big thing about it is we'll become 20% more efficient or whatever. or pick a number. I think the big thing about it is it's going to reshuffle the entire global economy in ways that have not yet been invented.

Will it change the way we need cognitive skills? Will our cognitive skills be changed?

I I would say probably. I mean, I think what's really going to impact things in the next five years is going to be quantum computing and that's going to take things and make it so much faster in everything that we do. it's going to wow I I don't need to go study that I know it already you know so what can you do with that knowledge and can you turn it into something that's very profitable for you or very enlightening for you I think that's what's really going to happen

how will we even interact with the three of us before we meet you rely on an AI agent to filter out our personalities and uh relations before we interact how humans interact in the age of AI.

But we're going to interact differently just like social media has made us all interact differently. Just like cell phones has made us all interact differently. No one even remembers how we used to interact before. You know, you could just pick up your mobile phone and send a text message.

Yeah, of course.

Right. And on that note, let us uh wrap it up. Please join me in thanking David Seagull and William. Thank you so much.