AI Is Reshaping Financial Wellness, but Can It Improve Retirement Savings?

Art by Marc Rosenthal

Since ChatGPT came onto the scene in November 2022, there has been an explosion of public interest in generative artificial intelligence that can use large troves of data to create almost instant text, images and other responsive content at the click of a mouse.

For those close to the AI industry, this technology has been in use and advancing for years. That includes the financial services space, where financial wellness providers have been working to leverage the technology to provide personalized output to mass audiences. Take, for example, Morningstar Inc.’s Mo adviser, an AI-based chatbot launched in April intended to respond to investor inquiries.

Defined contribution retirement savings, with plan advisers focused on engaging millions of participants at a time on both financial wellness and wealth management, seems particularly fertile ground for leveraging generative AI capabilities. Dani Fava, group head of product and innovation at financial technology firm Envestnet, is leading a team working on just such uses of AI for the financial advisement community.

Fava, who has a background in both technology and wealth management, says there are two key advantages to generative AI in the financial services sector. First, the technology allows those programming it to cater the use of the data to their needs.

“When you prompt a large language model, you can prompt it to be mindful of events and needs,” she says. “For example, if I don’t want to give investment guidance within an answer, the large language model can be trained to avoid that.”

Once those parameters are set, she says, the second advantage is that the generative AI can start making connections with a client’s various data inputs on its own, at rapid speed.

“You start to envision, ‘Can I connect the dots?’” Fava says. “‘What do I know about this person?’ I know their age, their demographic, maybe I know some other preferences that this person has answered through a questionnaire. … If I have these various pieces of information, then generative AI is really great at tailoring communications from both a marketing and a customer service perspective.”

While Envestnet is taking things slowly with client-facing capabilities, the ultimate goal is to use generative AI to help advisers engage more clients at scale, which then generates more data collection, which produces more opportunities to provide services.

But it’s not just about client-facing interactions, Fava notes. Generative AI can also be a tool for advisers when managing things such as retirement plan investment lineups.

“Think about how we create the best plan lineup,” Fava says. “It’s by looking at all of the potential holdings or funds that we can include in the retirement plan, and then all of that analysis, comparative research and data analytics that the adviser has to have in front of her. Generative AI can process through that a lot faster and can present the plan adviser with some ideas.”

Fava gives the example of loading up a company 401(k) plan, then comparing it to every asset class that Morningstar data has reported over the last five years. That analysis can be done by AI at rapid speed to potentially identify useful changes or considerations for the plan lineup.

“Generative AI can understand the context of the question and quickly add the new information you just gave it,” she says. “It’s similar to having a conversation with a really knowledgeable person, but the AI can do it in seconds.”

The Human Element

Jay Jumper, CEO of Future Capital, leads a firm that provides a financial wellness and managed accounts platform to advisers, connecting them to clients’ workplace retirement plans. He says AI helps personalize financial wellness interactions and prompts for participants, so long as the data is available to do so. The issues, he notes, is that data being available to advisers in safe and secure formats.

“It’s all about the data,” Jumper says. “Who has it and who controls it.”

Recordkeepers now have large troves of participant data that, if leveraged by AI tools, can identify financial wellness solutions and patterns for large pools of participants, Jumper says. There are currently two challenges, however, in putting that data to use.

One is getting the data to be shared, safely and securely, between recordkeepers and financial advisers. In some cases, the retirement plan providers are guarding that information for the potential to generate their own wealth management fees. The second, Jumper notes, is a fear that if data is pared with AI tools, it will replace humans with the development of robo-advisers or self-serve financial wellness tools. That is a future, however, about which he is not concerned.

“In our opinion, nothing is going to replace the relationship to the adviser,” Jumper says. “AI is a really great tool, and what we can do five years from now will be even better than what we can do now. But you are not going to replace an adviser that you know and trust.”

Jumper believes that, when a participant is ready to make a major life decision—which may even be be prompted by smart AI—they are going to seek human advice and counsel.

“The AI can guide you on the journey,” he says. “But when you get to making a decision, you’re going to want to talk to someone.”

Jumper believes AI will continue to increase its share of client interaction and engagement. But rather than replacing human advisers, it will help them reach more clients.

“The tool makes financial wellness more scalable,” he says. “But it also makes the advisers become more scalable and [more able to] go downstream [in their client services.]”

Slow and Steady

Technology firm Lucidworks, which uses large datasets to create insight-driven applications for businesses, surveyed more than 3,000 companies about how they are using, or plan to use, AI. Lucidworks found that investment in AI is burgeoning in sectors including financial services, with more than 90% of firms putting funding toward the technology.

But while interest and investing are happening, the financial services industry is moving somewhat more slowly than sectors such as technology and consumer products, in which generative AI is being used to drive revenue and business growth, according to Lucidworks CEO Mike Sinoway.

“More so than any other sector, with the exception perhaps of government, financial services companies are scared,” Sinoway says. “They’re scared of making a mistake that causes them to have a security breach, that causes them to give a wrong answer to somebody or give a response that is insulting or offensive.”

Sinoway says financial services companies absolutely want to engage in AI and use it to drive revenue. But unlike some other sectors, financial firms are investing in compliance for areas such as client service and guidance, as much as the actual technology implementation.

“In financial services, it’s more about customer service, more about cost reduction and much less about growth and revenue enhancement right now,” he says.

As an example, Sinoway says Lucidworks has a client in retirement planning that wanted to use generative AI to respond to customer inquiries in a chat environment. One way Lucidworks approached the request was to build a large language model that could access that client’s internal documentation and materials. Another approach it explored was accessing the much larger, public pool of data from ChatGPT, but that second path, though tempting, would also be more risky.

“If a customer asks a question like, ‘How do I open a 401(k)?’,” there’s no guarantee you are not going to pull a competitor’s write-up on how you do that or, worse, something that refers to one of your competitor products,” he says. “That’s the challenge. The best stuff is the riskiest stuff.”

Sinoway says Lucidworks ended up using internal data only, as that provided enough material and potential to “move the needle forward.”

“That’s the decision process and calculus all of these companies are going through right now,” he says.

Reaching a Wider Pool

Sean Murray, the head of retirement at Envestnet, believes that, if done right, AI can continue to drive more financial guidance and advice downstream to participants who may not have enough assets to get one-on-one adviser attention.

He notes that managed accounts, which participants often have access to within their retirement account, is already using technology to connect with and help manage client assets.

“Managed accounts is personalization for all,” he says. “What they are doing is, in the most simple terms, having a participant do things in a digital way without having to touch a human—and the reason you don’t want to have to touch a human is that it’s expensive.”

One of the issues with managed accounts, right now, is that even if participants choose to go into them, or are defaulted, they still need to engage to get the maximum benefits, Murray notes. But even here, AI can play a role by helping to nudge participants along with personalized prompts, he says.

If, in fact, participants can be prompted to give their full financial and wealth picture, then managed accounts can start to operate more like a target-date fund that, rather than just adjusting to age, can adjust to all the various pieces of data for a participant.

“I think that integration piece is where a lot more can be done,” he says. “The more information the providers get, and the more engaged the participant comes in, that’s where this thing really starts cooking.”