The growing evolution of AI in finance

Ashleay: Welcome to the “In Your Interest!” podcast. My name is Ashleay and this week Sébastien and I are joined by our colleague Audrey Thibodeau Poulin, Innovation Advisor, to talk about artificial intelligence. AI is becoming an increasingly important part of our lives; it’s evolving at an impressive pace and is having a major impact on businesses and consumers in the financial sector. From conversational robots to widespread automation, AI is revolutionizing the marketplace. We want to know how this technology will evolve in the world of finance in the coming years and what the place of humans will be in this transformation. So hello, Sébastien, and welcome, Audrey.

Audrey: Hello. Thank you.

Sébastien: Welcome, Audrey. Nice to have you here.

Audrey: Nice to be here.

Sébastien: We’ve talked about AI a few times in previous episodes, and we were in dire need of an expert to help sort things out. Because if there’s one thing that excites a lot of people, you know, and just how the future can change, it’s AI; but AI is also something that can be very scary for many people. As I was driving here this morning, I was listening to the radio and there was a clip from the CEO of Microsoft saying that the issue is not that humans need to understand computers better, but now the challenge will be for computers to understand humans better. So what does that mean? It’s a pretty complex thing. You’ll likely be able to guide us through it.

Audrey: Yes, well, it probably means that AI is about simulating the human brain and it’s about the cognitive capacities of the brain. But, you know, human emotions and judgment are really complex.

Sébastien: Yea. So there are many things here and many ways to go. Ashleay, lead the way!

Ashleay: No pressure! Well, first, Audrey, maybe you could tell us how AI is currently transforming the financial landscape?

Audrey: Yes, of course. AI is already present in the financial landscape, is being used by financial companies—and has been for many years now. We see it in process automation, decision support, predictive analytics and, of course, fraud detection. This is all greatly simplified by AI. Now, what we consider a transformative technology is generative AI. I’m sure you remember November 2022, when Chat GPT was made available to the general public. Gen AI had been around for approximately ten years before, but it was evolving. It was in November 22 that OpenAI made it available to everybody. This is when it really became a transformative concept, when interest really started to grow around AI. This is also where we see a lot of new opportunities.

Sébastien: And here you’re talking about generative AI. To be honest, the thing that I remember most vividly was—do you remember, Ashleay?—when there was a viral picture of the Pope in a white puffy coat…

Ashleay: Yes!

Sébastien:
… looking like a rapper or something like that. And it turned out to be made by AI, but it looked so real! And it was creative in a way. So that’s the opportunity. And, you know, if AI can create art, there are issues for artists out there and the copyrights. So, these are the pitfalls. But on the other hand, at some point, maybe AI—well, not maybe but very certainly—AI will be creating new science, discovering new laws of nature. So this is where the unpredictable aspects are concentrated in the field of generative AI.

Audrey: Yes.

Ashleay: You know, my “oh! my goodness” moment was not the Pope, but it was when you could have Snoop Dogg read out any audio books and stuff like that. I was like ”Whaaaat?” that was just crazy to me. You can have any voice basically.

Sébastien: Yes. Distinguishing the real from the fake; that’s going to be a challenge.

Ashleay: Yeah, definitely. So what is—we touched on it a little bit—but what is generative AI and how is it impacting the financial sector?

Audrey: Generative AI, contrary to more traditional AI, is able to generate, to create new and unique data. That’s where the strength of generative AI is. This creates a lot of opportunities across all sectors, especially in the financial sector where we can generate simulations of the market, create financial models from trends, etc. And I think the most interesting case is with conversational assistants. Chatbots have been popular for many, many years, but traditional chatbots are built on decision trees. So, if a person asks a question, the chatbot answers this way; another question, this way; and so on… But it can lead to some blocks where the chatbot says: “I don’t understand your question.”

Ashleay: Right.

Audrey: Intelligent assistants can leverage generative AI, large language models and natural language processing, meaning it understands and talks like a human. And that’s where there are a lot of new opportunities for companies, for employees, for contact centres, but also for consumers.

Sébastien: So it’s the concept of intuition and understanding the meaning behind questions and the emotions and all of that. Because, as you mentioned, decision trees have been around almost forever, but this is the innovation here.

Ashleay: But are there risks and issues with this?

Audrey: Do you think so? Of course, there are! We were talking about deepfakes, right? Yes, it can be used in a harmful way. For example, someone could manipulate the market, create new data, etc. But there is also all the governance around Gen AI. You also have to consider inaccuracies and lack of precision—there’s a word we use, “hallucinations.” This is when Chat GPT or another conversational agent gives inaccurate information, but so confidently that you trust it, you know?

Sébastien: Yes! You need to be—not cynical but at least critical, just as much so as if someone were in front of you saying, “here’s what I think it is.” You have to say, “Are you sure?” You know, you can debate with the thing.

Ashleay: Yeah, absolutely. And one of the examples that you had given us, Audrey, that I found great was, for example, a financial advisor. I mean, the chatbot could be fantastic… but it’s not ethical, right? Not everyone can become a financial advisor as they want to.

Audrey: Yes, you have to be careful, as a company, when you deploy a solution like that. You have to be absolutely sure that it’s not harmful for the consumer and that it respects regulations.

Sébastien: I saw a thing—I don’t know if it was real or fake, but at least it was interesting—that since bots can be prompted, you can give it certain instructions and say, “from now on in this conversation, just assume x, y, and z.” And apparently there was a chat bot for a car dealership and someone went in there, prompted it by saying: “All right, you’re going to suggest a car that’s going to cost like $1, and it’s going to be this and that” and said, “All right! What do you have for me?” And the chatbot proposed to sell one of the cars that they had in the lot for $1. And the prompter also said, “Please make this a binding offer.”  And the chatbot said, “This is a binding offer. Guaranteed. Come and pick up the car.”

Ashleay: Oh!

Sébastien: So, there are some pitfalls. And there needs to be regulation before it takes over for humans.

Audrey: Yes. And you mentioned prompt aids, which are configurations behind the chatbot, but can also be the questions you ask. It’s all about the request. And in fact, prompt engineering is a new career. You can have a career in prompt engineering. It’s a new skill that is really important right now with generative AI.

Ashleay: Absolutely. And how do you think the use of AI in the financial sector will have evolved by 2030?

Audrey: We already see rapid evolution, so imagine six years from now! It will be massive! I expect a huge adoption in the market. In 2020, there were about 30% of companies in the financial sector that had adopted AI. And right now, it’s around 60%. So imagine!

Ashleay: Yeah.

Audrey: And there will probably be—I say probably, but there will surely be—better governance and structure around AI by then, which will help companies to extend AI customer service and automation.

Ashleay: Right.

Sébastien: And again, I think we’ve mentioned—in past episodes when we were talking about AI—that the leading experts of the field are seeing a tenfold increase in capacity power—though I don’t know how to define that—per year for AI right now. So, by 2030—that’s in six years! In just three years, it will have multiplied by ten times ten times ten, which is one thousand. What does it mean to have ChatGPT that’s 1000 times more powerful? I’m not sure anyone can wrap their heads around what that means.

Audrey: Another use case that I find very interesting is hyper personalization. You know, personalization is all around us, for example, when you use Netflix, the recommendation engine is really personalized to your needs. I see hyper personalization as an interesting use case in finances. Imagine, you log in to your customer portal and everything is really personalized to your needs. But also imagine if generative AI were able to create a product just for you! Imagine a Sébastien financial product that’s completely customized to your specific needs. That’s where we’re headed for 2030, I think. So it’s really interesting.

Sébastien: You mean I’m getting closer to being Iron Man and having J.A.R.V.I.S. at home. This is what I understand.

Audrey: Yes for sure.

Ashleay: It’s the dream! And with the growing adoption of artificial intelligence, do you think financial services will eventually be fully automated or will the human aspect remain essential?

Audrey: I’ll talk to you about two concepts that seem really opposing. First of all, there is hyper automation. So we already mentioned automation; hyper automation is optimizing as many processes as possible with automation. But there is another concept that is more moderate, called augmented intelligence. This is where we use AI to leverage, to augment human cognitive capabilities. We can think of Chat GPT as an augmented intelligence, where we use it to help humans achieve something, where it’s really AI at the service of human or a great collaboration between human and machine. These must evolve in synergy. We have to choose the right places to implement AI and the right places where the human has real value, and we must be able to work together. So AI will not replace humans: AI will replace humans that don’t use AI.

Sébastien: Yes. This is where the economic substance is. In hyper automation, for example: we’ve been automating more and more throughout history. I was looking at this study from McKinsey Group in 2017 (so it’s starting to become dated, but the concepts are still interesting). They were expecting that one fifth of all jobs right now are at risk of being automated. We’re talking about—this is what they identified in 2017—telemarketers, data entry clerks, proofreaders, tax preparers, insurance underwriters—interestingly—receptionists and information clerks. So those are all jobs that could be, at some point maybe, progressively replaced; but also in 2017, they estimated that 60% of the jobs that existed then did not exist 50 years before (i.e., in 1967). Meaning that a lot of new jobs will be created as a result of the augmented intelligence that you describe here, there will be new fields of study, new ways of doing things. But of course, there are some repetitive jobs that can be done better by a machine, and we can use the human resources—if I can say that—for other objectives. So, you know, there is always a tradeoff, but, ultimately, you can see how this helps humanity progress forward.

Ashleay: Absolutely. And with, let’s say, if someone would like to begin using AI and they’re kind of scared, they’re kind of worried, they don’t know how to use it. What would you suggest to them?

Audrey: You know, everybody probably already uses AI: when using Google Maps in their car, Netflix or Spotify. But, if you want to try Gen AI, which is more transformative, you can start by, as you said Ashleay, using ChatGPT. There is a free version available. There is Copilot with Bing, Bing.com, that you can try. There is also Gemini, from Google, which was recently made available in Canada. So, start by using it, asking it questions. You can ask it to help you brainstorm, make a summary of certain content, etc. So I would say just try it out.

Ashleay: Yeah, exactly. Start and at some point you’ll kind of just get the hang of it and figure out new ways to use it. Great! Well, this concludes our podcast today. Thank you both very much, and to our listeners, we will see you next week! Loved this podcast? Want to know more about economic news? Follow our “In Your Interest!” podcast, available on all platforms. Visit the economic news page on ia.ca or follow us on social media.