Financial planning educational departments are investing in “role-playing” artificial intelligence agents to help aspiring advisors learn how to interact with different types of clients. Think of AI-created personas able to respond to, and ask questions of, a financial advisor, and then evaluate the quality of that advisor’s side of the interaction.
It’s a mostly positive use case for AI in financial services, and one in which academic institutions seem to be ahead of the advisory firms. Many firm leaders are looking to balance the beneficial impact of AI tools on training and operations, with the potential negative impact of becoming too reliant on the tools to the point where new employees won’t have the same real-world, on-the-job opportunities to understand the complexities of running a high-touch service business and to develop the empathy and emotional intelligence required.
These were points raised during a “Focused on the Future” panel discussion, held during the executive sessions prior to the Wealth Management Industry Awards gala in New York last week. Firm representatives, college educators and advocates came together to discuss the impact of AI on a firm’s employees and trainees.
Suzanne Siracuse, the discussion’s moderator and CEO of Suzanne Siracuse Consulting, asked the firm leaders if they had in-house AI-enabled role-playing agents to train new advisors. Most said they did not—though they were interested and a few had made tentative steps in that direction.
In the discussion, Shaping Wealth CEO Brian Portnoy noted that AI’s presence in the public space was still new, reminding the group that ChatGPT only launched for public use in late 2022 (compared with the general use of the Internet, which grew over decades).
To Portnoy, AI will increasingly commoditize the technical elements of advice, turning 10-hour tasks into 1-hour ones, 1-hour tasks into 10-minute ones, and making 10-minute tasks disappear altogether. But nascent advisors may lose opportunities to develop skills and familiarity with the industry as these tasks dissipate.
George Nichols, the president and CEO of the American College of Financial Services, agreed AI could create more opportunities to train advisors by simulating the scenarios they might encounter during client conversations.
“I hope we can create a simulator no different than we create a simulator for a 747. I would never put you in a real 747 and let you practice,” he said. “But I’ll let you in a simulator, and you can crash it every single day. But then I’m going to be able to teach you every single day about what you’re doing right and wrong.”
Portnoy added that the most popular tool in Shaping Wealth’s “Lydia” (the firm’s behavioral wealth assistant) is a role-playing agent. The tool can create simple or randomly generated scenarios for advisors of any skill level to practice. Portnoy said the agent was even being used by a senior executive at a large firm to test alternative approaches to more complex clients and situations.
One of the features, Portnoy said, is that the tool doesn’t tell users what an advisor should say, but gives feedback on where the advisor could have improved and asked questions differently.
“It doesn’t have images of people, but it does indicate body language,” he said. “And if you didn’t pick up on the fact that he’s being aggressive and she’s slouched over and it’s a difficult conversation … Lydia’s going to say you didn’t pick up on that. And maybe the next time around you can do a little bit better.”
Craig Lemoine, the director of the financial planning program at the University of Illinois at Urbana-Champaign, agreed AI agents offered an exciting method to train industry recruits.
Lemoine said he’d struggled for decades to recreate client experiences for students, using a variety of options, each with its own shortfalls. Hiring actors to play the part of the client, or using video and audio with multiple scenarios depending on what responses the trainee chooses, are options, but the range of learning tended to be static, he said.
His university program has created two agents for role-playing purposes, including “Lucy,” in the guise of a recent widow in a financial crisis. Students can upload outputs to the agent and interact with Lucy, who’s been constructed on the Gemini platform. At the end of the interaction, Lucy creates a rubric detailing how the student performed (even noting whether the student offered Lucy any refreshments).
For Kristy Archuleta, a professor of financial therapy and financial planning at the University of Georgia, the AI debate underscored broader concerns about the so-called “soft” skills of human interaction, communication and empathy. She made it clear that even with AI training, students would never know exactly what kind of client or complicated family situation they might encounter in the real world, and developing those soft skills among actual human beings was important.
She said while the AI tools were useful, she was concerned about educational pathways that don’t require students to learn more about how to talk to clients.
“How are we ethically doing our job as fiduciaries to be able to talk to our clients when we’re not training our students to do that?” she asked.
According to panelists, some students (and prospective planners) are being put “in a bit of a bind” between their uncertainty and, at times, adversity to AI, and the firms forging ahead with an “all-in” approach to the technology.
The college world’s “blowback” to AI in just the past year has been “huge,” according to Jamie Hopkins, the chief wealth officer and CEO of Bryn Mawr Trust Advisors. Hopkins said students were excited about AI a few years ago, but now more are hesitant about pursuing careers with companies that rely on the technology.
“They are starting to say, ‘I don’t want to be at places that are saying we’re going to be AI. I want to go be a person,’” he said. “I think probably everybody here has some signs around in your town that say ‘no data centers here.’ They’re still popping up, and that’s a very interesting fight.”