JPMorgan Chase’s Ozzie Solares on the tension between AI hype and reality

Ozzie Solares, managing director of asset and wealth management at JPMorgan Chase, thinks AI’s impact on marketing is undeniable. But he’s also realistic in its practical uses.

“My challenge with it is, when you look at AI, it almost looks like magic,” he told The Current’s Chris Brooklier. “Anything that that machine spits out seems real, and then what you ultimately experience as a practitioner oftentimes is that it’s something less than perfect.”

In a conversation at Possible in Miami, Solares talked about the AI hype cycle, how to know when to place big AI bets and why the technology can’t make bad marketers good.

How is your team at JPMorgan Chase integrating AI?

It goes without saying this is a fundamental shift in the world today, as far as how people work and how technology is getting integrated. It’s not only impacting marketing but every aspect underneath it, whether it’s finance, operations, sales.

AI will be an enabler for the humans that are there. Humans with AI are going to be ultimately superior to a human on their own. So the question that’s being asked is: How can we maximize that? In marketing, that’s looking at how to automate workflows, how to free up people to do more strategic work. But again, we’re still in early phases, so exploring the full potential.

What is the tension between the AI hype cycle and its practical uses?

Having to show shareholder value, having to show a return on something like an AI investment, right? There’s a lot of pressure on us to be able to make advances in that. When you’re sitting on the front lines, you feel the whole brunt of that coming in the form of: trying to bring in the tools, trying to show the actual productivity gains, trying to keep morale where it needs to be because these changes are happening over time.

My challenge with it is, when you look at AI, it almost looks like magic, especially when you’re not on the front lines actually doing the work. Anything that that machine spits out seems real, and then what you ultimately experience as a practitioner oftentimes is that it’s something less than perfect. That isn’t to say there isn’t promise there and that it can’t eventually get to something that’s worthwhile. But my opinion, in a lot of what we’re seeing, is that we’re not quite there yet.

Everything is changing so fast. How do you know when to place a big bet?

It’s a balancing act. There’s pressure on you to move quickly, so in this process, you’ve got to assume you’re going to place a few bad bets. The question is how to hedge in that scenario. One is [that] you try to identify the areas where you’re going to have the most impact and see the most short-term gains. The second is where there’s the least amount of risk. At a company like JPMorgan Chase, a lot of that comes down to financial regulations and the risks associated with that. It’s in that balancing act that you eventually prioritize what you’re going to do.

How do you think the dot-com bubble compares to the AI era?

It definitely feels familiar. I think the main difference is the magnitude of the impact. This is a technology that’s impacting every single industry, every single job that’s out there. To even try to wrap your head around what the world is going to look like two, three, four years from now is a fool’s errand, because the reality depends on how quickly the technology gets embraced. Nobody knows that. … The ability to enable and provide marketers [with] literal superpowers in what they do is incredibly promising.

Can AI make a bad marketer good or a good marketer great?

It cannot make a bad marketer good. Without being a subject matter expert, using a tool that’s just spitting out information without being able to discern the quality, is absolutely useless. Couple that with the nature of AI and LLMs and how they work, they’re prompt-based systems. So your outputs are only as good as the inputs. That leads me to believe that if you’re a good marketer, you’re going to have good inputs and you’re going to get better outputs, and that’s going to lead to a better use of the technology.