English transcript:
00:00:00:02 - 00:00:20:19
Grant McDonald
2025 felt like a turning point for artificial intelligence. A year when AI truly broke through and became part of everyday life. From tools that save time and automate tasks to innovations that reshape entire industries. The pace of change has been remarkable. But what does that mean for the future of banking and work in general? I'm your host, Grant MacDonald.
00:00:20:19 - 00:00:43:18
Grant McDonald
And today on the Banking on AI Podcast, presented by TD Invent, we're joined by Jeff Chan, Senior Manager, Enterprise Innovation here at TD, who spends his time looking beyond the headlines to explore where we're heading. And AI is a very big part of that.
00:00:43:20 - 00:01:02:10
Grant McDonald
Jeff, thanks so much for being here. I really appreciate it. I'm always intrigued by the work that you do at TD. I feel like this year was a big year for AI. Like 2025 was a year that it hit mainstream and just went at the speed of light. I'm very curious as to how you see it.
00:01:02:10 - 00:01:06:11
Grant McDonald
Your take on that. Like, do you agree with that? This was a year for AI.
00:01:06:13 - 00:01:26:15
Jeff Chan
It was definitely a year for AI, but I think AI's gone through a couple of mainstream moments. I mean, AI is basically another component of the automation journey that the world's been on for, for centuries now. You know, I started almost 50, 60 years ago. It's had its ups and downs. It's had a couple of ups even prior to this.
00:01:26:17 - 00:01:59:14
Jeff Chan
What makes it feel so mainstream this year, actually, in the last two or three years, is because that little company that did what they did with that chatbot, it did three things pretty well. You can talk to it. It seemingly generates brand-new content and it's relatively cheap to deploy. And when I say cheap, I mean before this thing came out, someone like a TD would have spent seven, eight figures to train something like this to put in front of their customers or colleagues.
00:01:59:16 - 00:02:16:10
Jeff Chan
And now it's much cheaper. So I think that company, ChatGPT, coming out, it did those three things. And suddenly you don't need a computer science degree. You don't need an engineering degree. Suddenly, it's in the palm of your hands. As long as you can conversate with the thing, off you go.
00:02:16:12 - 00:02:28:07
Grant McDonald
So Jeff, in the little intro that I do for this podcast, I talked a little bit about you. What you do here? I think I got it pretty, pretty accurate. But I'm always curious as to how you describe what you do here at TD.
00:02:28:09 - 00:02:46:06
Jeff Chan
Yeah, I'm very fortunate at TD. So, my role is basically, I try to figure out what consumers want next, and why. So, I look at the intersection between shifts in emerging tech and consumer behavior. If I know where it's coming together then I can help the Bank reverse engineer what it means for products, services and technology.
00:02:46:08 - 00:02:58:21
Grant McDonald
That is extremely interesting. You know, when we're thinking about what TD is doing and how it's doing right now, when we look at AI in the year that we've seen, how would you characterize it? How are we doing?
00:02:58:23 - 00:03:16:01
Jeff Chan
Yeah, I think, TD we're right in the middle. So it's not good, it's not that bad, we're right in the middle. And I think just like all banks, we're focused on productivity, efficiency, basically augmenting how our colleagues work today. All the all the other banks are basically chasing the same thing right now.
00:03:16:03 - 00:03:26:11
Grant McDonald
When you're having conversations with individuals here and you're saying, okay, we're in the middle. What do you, what are you talking about in terms of where we could be going to move a little bit? You know, in the right direction.
00:03:26:13 - 00:03:50:00
Jeff Chan
And I think the good news is we, we are. That's the good news. Where I see the industry going with this technology is instead of generating content, we're actually training the AI to take action on things. And that's where we're going to start to see a lot of gains. We're going to start to see a lot of our internal processes being automated by this technology.
00:03:50:01 - 00:04:18:04
Jeff Chan
So that you and I can focus on the more value-added opportunities that may be more complex or may be too messy for an AI to frankly, figure out. Because no matter where you work, processes are always messy, and if the process is messy an AI will likely fumble as much as a human does. So, there's moments where the AI will come in and automate stuff, complete actions, and then there's moments where we will come in and do what we need to do, and also supervise that AI.
00:04:18:06 - 00:04:38:15
Grant McDonald
A lot of what you do, you know, when you're looking at these different parts of consumerism and how people are reacting, what they're doing, you know, 800 million users of ChatGPT weekly, that's a significant number. So, you look at something like that. How are your feelings around adoption rate and where it's being used, where we can really start leaning in a little bit more?
00:04:38:15 - 00:05:03:21
Jeff Chan
Yeah. So I get this question a lot. I always say, AI has always been here. The funny thing about AI is we tend to move the goalposts mentally. As a society, what is classified as AI? If I can explain it, suddenly it's not AI. If I can't explain it, it's suddenly magical and it's AI. So, yeah. ChatGPT 800 million users.
00:05:03:22 - 00:05:21:23
Jeff Chan
But if it wasn't for AI, Google wouldn't be around. If it wasn't for AI, Amazon couldn't tell us what to buy next. Netflix couldn't tell us what to watch next. Tinder couldn't tell us who to date next. Som The Weather Network, we know when rain would stop down to the nearest minute, right? So, in terms of adoption, it's always been there.
00:05:22:00 - 00:05:39:07
Jeff Chan
It's just that again, with ChatGPT, it's suddenly in the forefront. It was always playing this, like, behind the scenes magic thing. And now it's at the forefront, as long as you can talk to it, it seemingly can generate or take action on your behalf. And that's what's so magical about ChatGPT.
00:05:39:09 - 00:06:03:16
Grant McDonald
When you look back at the past 12 months or so, I mean, even before that, people always have the November 2022 date in their mind. They just, there was a significant moment. I'm curious, from all the research that you've been doing this year, do you have some key moments that you see? Okay. That's interesting. I think that's going to have an impact and not to your point, like specifically AI, but all things around it.
00:06:03:21 - 00:06:05:18
Grant McDonald
Had there been moments that stood out to you?
00:06:05:20 - 00:06:26:23
Jeff Chan
Yeah. Looking back, things like, when the term was coined in the 50s to our own Canadian Dr. Geoffrey Hinton, doing what he did with neural networks and, and advocating for that even when AI had that period, we call an AI winter when people weren't focusing on it. But he just kept pushing because it wasn't enough compute power.
00:06:26:23 - 00:06:46:20
Jeff Chan
There was enough horsepower to bring to life what he wanted to do. And then suddenly, about ten, 20 years ago, suddenly we got the horsepower we needed to bring his research to life. And now he's celebrated, right? So that's definitely one. ChatGPT is a huge moment again, because we broke through that barrier of it was difficult to talk to a machine.
00:06:46:22 - 00:07:15:22
Jeff Chan
Now it's not. And if I look ahead, for me what's interesting is not AI by itself. It's when I collides with other disruptions. So, AI with blockchain, AI with spatial computing, AI with robotics, AI with nuclear fusion, AI with quantum computing. When these things start coming together, that's when things get, I won't say more exciting, but "more exciting."
00:07:16:00 - 00:07:35:17
Grant McDonald
I'm not going to go into, like, looking into a crystal ball. I mean, I might later, but I am very curious about the conversations that you have. I'm sure people, question number one, what are the risks associated with this? What's sort of your response to that? What's your go to in terms of saying we, we need to lean into this, but we can be careful as well.
00:07:35:19 - 00:08:02:17
Jeff Chan
Yeah. So, about ten years ago, we started this era of move fast and break things ten, 20 years ago. And that term was popularized by Facebook, frankly, move fast and break things. Zuckerberg used to have posters that said that on campus. Well, we've passed that era. We're now in the era of slowdown and fix things. And that's why when we talk about this topic, even the questions you're asking me today, they're not.
00:08:02:17 - 00:08:27:09
Jeff Chan
So, oh my God, what could we build? It's what should we build? And what questions should we ask ourselves? So, when I have the opportunity to talk to colleagues, students, even my own parents, I always say we should all keep asking these questions because we have to slow down. We have to fix things. The only industry that should be moving, in my opinion, the only industry that should be moving fast and breaking things, are government.
00:08:27:11 - 00:08:45:10
Jeff Chan
They need to catch up. They need to catch up with the rules. You're likely going to hopefully see more countries coming up with safety, privacy laws that govern and protect both you and I and the companies that do business with us. We're going to see countries come together in a more coordinated fashion to create those rules, and you'll see them updated constantly.
00:08:45:10 - 00:08:51:00
Jeff Chan
And that's what they need to do now. It's, it's the government's turn to update these laws as quickly as possible.
00:08:51:02 - 00:09:15:07
Grant McDonald
It's I mean, you've hit it right on in terms of the policy being, it is naturally designed to be slower moving, and now it's up against something that moves extremely fast. When we look at, you touched on it already. We can now talk to, essentially a computer. What other moments do you see where it became easier for consumers to actually interact with this stuff and not necessarily be afraid of it?
00:09:15:09 - 00:09:37:00
Jeff Chan
Yeah. So probably would start in like the 80s with the little prompt. And then in the 90s when the Apples and the Microsofts of the world created the mouse interface, the GUI interface that made things more practical. Again, my background's computer science. So we've been on this journey of trying to create machines we could talk to basically as easy as possible.
00:09:37:02 - 00:09:44:05
Jeff Chan
And the easiest way for humans to communicate is through language. And now we can use language, everyday language, to talk to a machine.
00:09:44:07 - 00:10:03:06
Grant McDonald
I love hearing that example because I mean, essentially, like the game solitaire and a few others are put on the computer so people can start using the mouse properly, like if I recall. And it's those little cues that you see. I'm, I'm curious in terms of how the perception has shifted when you chat with people about AI. Like what?
00:10:03:08 - 00:10:06:04
Grant McDonald
What is the change that you're seeing or hearing out there?
00:10:06:06 - 00:10:39:15
Jeff Chan
Yeah, consumers, actually in general, people are as excited as they are concerned about this. And again, that's the era we live in now. And they ask a lot of questions around privacy. They ask a lot of questions around addiction. And you've seen OpenAI talk about a term called parasocial relationships with technology. And again, it didn't start with ChatGPT because we've seen this behavior in Asian countries where certain adults grow, a relationship with avatars, digital avatars.
00:10:39:17 - 00:11:12:12
Jeff Chan
And that started before now, this chatbot isn't just a visual representation of who they wanted to be with, it can actually talk back and forth. Right? So, there's a lot of concerns around, again, addiction quote unquote, unhealthy relationships, privacy, but there's also a lot of excitement because again, without AI, we likely would not have discovered the vaccine that would allow us to be in this room together five years after a brand-new virus was discovered.
00:11:12:14 - 00:11:24:14
Jeff Chan
Right? So, there's a lot of good that AI can bring, and there's also a lot of bad. I never want to blame the technology. Technology's not bad. It's the person who uses it that could make it good or bad.
00:11:24:15 - 00:11:42:13
Grant McDonald
You know, it's a very good point in terms of that, the speed behind it. And at an institution like ours, at TD, we are a very large organization. We have a goal of moving faster, a bit simpler. What industries are you seeing doing a pretty good job, incorporating AI into their day-to-day?
00:11:42:15 - 00:12:07:13
Jeff Chan
Yeah, I see like any industry that wants a competitive edge will use it. So, retail. Before TD, I was in retail, I was in commerce. And they use it, a lot, for personalization. Again, the next best action to take, or the next item, or the next movie you should watch. So, you see a lot of AI there. I'm even seeing in entertainment, you're starting to see them use it for creative development.
00:12:07:15 - 00:12:23:20
Jeff Chan
Netflix, for instance, they're using AI to do virtual dubbing of actors and actresses. So, you can speak in English in a movie, and the AI will literally dub, not just the sound and tone of your voice, but your facial movements, too. And that's already on Netflix.
00:12:23:21 - 00:12:43:12
Grant McDonald
So, when you look at all of the things that you see in your day-to-day, the stuff that we talked about today, what's the main message that you're trying to leave people, or is there one or is there several where you're saying, here's what you really need to be thinking about over the next year? Because I see this is just it's just going to keep getting faster.
00:12:43:14 - 00:13:09:03
Jeff Chan
Yeah. I always say, don't think of AI. It's just a tool. Innovation is about solving a problem in a new or different way to generate some value for that end user. So, focus on the problem. If you focus on the problem, and then in your mind, wave a magic wand of what idea could solve that problem, and then think about the solution itself, like if it, if it happens to be AI.
00:13:09:05 - 00:13:24:15
Jeff Chan
If people are focused on AI, I always say, just be mindful of how you're using it. We should continue to ask ourselves these questions so that we build a system that is very transparent, ethical, responsible, and democratized for everyone, anyone and everyone to use.
00:13:24:20 - 00:13:39:23
Grant McDonald
Excellent. Now, you mentioned earlier we're asking the question, what should we build? But I have to ask because you're sitting here, what, what could we build? What could actually come up, a few different things. There's got to be a few standouts where you say, this is a crowd pleaser. I'm gonna let them know what I think.
00:13:40:01 - 00:14:09:17
Jeff Chan
I mean, so right now we're focused on AI as this thing locked behind a screen. In then the next phase of AI, if you want to generate the most value out of it, it needs to jump out there. It needs to be in real life. And the term that's popping up is called physical AI. So physical AI is the collision of three or four disruptors, mainly AI, robotics and advanced sensors.
00:14:09:22 - 00:14:30:08
Jeff Chan
And you and I have an advanced sensor. If we have an Apple Watch, that's an, that's an event sensor. So when these things collide, AI can actually take action in real life to complete anything we want, almost anything we want. An example of a physical AI thing today are self-driving cars. So, a self-driving car with robotics alone can't happen.
00:14:30:10 - 00:14:45:11
Jeff Chan
A self-driving car with AI alone can't happen or with advanced sensors alone can't happen. But when they all came together, suddenly we have self-driving cars. So for me, that's one of the next things that are that are coming that I'm, I'm both excited for and maybe a bit concerned.
00:14:45:13 - 00:14:57:17
Grant McDonald
But listen, I so appreciate you taking the time to sit down today. Thank you for answering my question about what could possibly happen in the future. It is, it is really interesting to hear, you know, all the insights that you do day to day. So thank you so much.
00:14:57:19 - 00:14:58:20
Jeff Chan
No problem. Thanks for having me.