Should Advisors Use AI to Make Investment Decisions?

“It’s not deterministic, meaning it’s going to give you the best probable answer at any given point in time, and you’re never guaranteed to get the same answer, given the same input,” he said. “In the investment decisions, in my book, it’s a big no-no. You cannot have a decision-making process that is probabilistic. It has to be deterministic, meaning if you have the same set of inputs, you should expect to get the same output. And AI is not that.”

AI can help a portfolio manager or an advisor reach a decision or provide information to help them make one, but it should not make the decision for them, Naidu said. With other use cases for AI, such as meeting notes and CRM, it’s okay to have some non-zero tolerance. “The room for error is zero when it comes to investment decisions.”

There are certain investment workflows for which AI is useful, he said.

“The task that it is really good at is explaining things, and what is happening in an account or in a portfolio or different things it should try or an advisor should try, or focusing or bringing the attention of the advisor, the portfolio manager to the areas that are deviating from the norm,” Naidu said.

Some advisors and portfolio managers are spending a lot of time on these tasks.

“AI is really helpful in helping the advisor and user to get to those things faster because it can really look at a lot of data and be able to process that and come to say, ‘Hey, of the thousand accounts that you have, these 20 seem to be abnormal,’” he added.

Naidu said AI could eventually reach a point where it’s doing investment management, but asks, “Is that what we want?” One of the biggest benefits of AI is getting a different opinion.

“If everybody’s coming to the same opinion, then you’re not going to get the benefit of investing there,” he said. “You’re just following everybody. Less value-add. It feels like at this point AI can do pretty much anything that you want to do, given enough time and resources, but I don’t know if we want that to happen with investment decisions.”

Ugur Hamaloglu, who leads the EY Americas Wealth & Asset Management Consulting practice, said analytical tools supported by generative AI that are accepted by the market can serve advisors very well and make their lives easier.

But he agreed that AI today is not ready to directly serve immediate investment decisions. Investment management is risky and still requires governance and human oversight.

One of the problems, he said, is that even with the right data, these models still tend to hallucinate.

For AI systems trained on this subject and equipped with the right data, they can be helpful for conducting simulation analysis and stress testing.

“I don’t see much risk because that is insight for the advisor to take action on, or there is insight for the advisor to train their clients because the financial advisor job is part knowledge, but part like being a teacher, educator,” Hamaloglu said. “But making buy, hold, sell decisions, specifically promising certain financial outcomes. … So trusting a computer system, doesn’t matter how good it is, without human supervision for those subjects, I don’t think it’s reasonable right now.”

“Five years from now, we may look back and say, ‘Actually, they’re doing pretty well. Right now, we don’t have enough empirical data to say ‘They’re doing a good job’ or not.”

Hamaloglu said there are other fertile areas where wealth managers can implement AI, such as client acquisition. The risk is not as high as in investment management, and it can accelerate advisor outreach.