Artificial intelligence seems to be everywhere, with seemingly endless applications. However, many in the retirement plan industry are still figuring out how to integrate it into their workflow.

In a conversation with PLANADVISER, ERISA attorney David Levine spoke about how he has used artificial intelligence to develop a tool for retirement plan administration and his predictions for what is likely to come in the future.

PLANADVISER: How do you see AI being integrated into the retirement plan and DC recordkeeping and investing space?

David Levine

Levine: Where AI is most prominent in the retirement plan market right now is in the area of education and advice. Not a day goes by that I don’t see a new or existing solution promoting its use of “AI.” What I find most exciting is less the “we have AI” message, because—much like prior generations of computer innovations—it is rapidly becoming a core, rather than unique, function, but how it is used for increased personalization for stakeholders in the retirement and investment industries.

PLANADVISER: What are the biggest misconceptions about AI in the retirement plan industry?

Levine: The first misconception is that AI “thinks.” It isn’t conscious. It takes skilled programmers and those with subject matter knowledge to maximize its value. The second misconception is that AI will replace everyone. It isn’t true. We’ve all become increasingly “cyborg-like” over the years. For example, when is the last time someone read a map? Most people just use a maps application on their phone. It doesn’t mean we still don’t need maps.

PLANADVISER: How is the current AI potential different from the computer-based models of the past that have been used to develop managed accounts or participant advice platforms?

Levine: Traditional computer models have been more linear in that ‘A’ leads to ‘B’ output. AI and large language model systems allow more granularity and shades of gray that can enhance personalization.

PLANADVISER: Can you tell us a little more about your product and how it leverages AI to solve a problem or inefficiency?

Levine: PlanPort is a web-based solution that is designed to address a key challenge in the retirement industry: reading, understanding and translating plan documents for a variety of purposes, including onboarding, mergers and acquisitions, and participant-level services. To read and understand a plan document can take a significant amount of time, and PlanPort helps cut the time involved to a fraction of historic manual practices so that human subject matter experts can maximize the value of their time.

Although there is a lot of talk about AI, at PlanPort, we use AI as one component of a complex, layered process that integrates detailed knowledge of how retirement plans and plan documents “work” with an AI system that provides a key step in translating these documents.

PLANADVISER: How do you train or maintain your AI systems to keep up with potential regulatory changes?

Levine: Because our system is a “universal translator” for plan documents, we focus on maintaining key plan provisions as part of our output. We do not train AI models on client documents, and they are not “the product” that could be sold or leak out between our business partners.

PLANADVISER: What areas of the retirement plan industry can expect the most change due to AI in the next two to five years?

Levine: Aside from the impact of PlanPort, which our team is confident about, education and advice are most likely to be impacted. The biggest area of regulation and restrictions is one of your prior questions—the training of models on client data. I already regularly see negotiations over the restrictions on the use of stakeholder data in training AI models, and I expect that to grow.

PLANADVISER: How will AI be used to augment or replace human work or interaction?

Levine: I see it as an augmentation tool. The debate for decades has been: Does technology replace or augment human work? AI is good in many things and will evolve, but it makes a great companion—not a replacement for human beings with their judgment, knowledge, expertise and social skills.

PLANADVISER: Are there any ethical or privacy issues you can envision in the application of AI in the retirement plan space? What should fiduciaries be aware of when considering tools that incorporate AI?

Levine: As noted above, in the privacy world, there has long been discussion about what happens to individual and corporate data when used by third parties, and legal contracting is already addressing that point. The more confidential the information, such as personally identifiable information, the more beneficial it can be for stakeholders to understand how information will be used, shared and monetized. However, courts have generally focused on privacy as a contract matter and not, despite some attempted claims, as a fiduciary breach matter, so this topic may be yet another part of the contract negotiation process.

David Levine, a lawyer by trade, is the founder of PlanPort and a principal in Groom Law Group, Chartered, the largest employee benefits law firm in the United States. No Groom or Groom client information or resources of any kind are used in PlanPort, and PlanPort does not provide legal advice of any kind.

Any opinions of the author do not necessarily reflect the stance of ISS STOXX or its affiliates.

More on this topic:

	Regulators Urged to Take Risk-Based Approach Toward AI
	AI and the Future of Participant Services
	Moving Toward AI, Both Ethically and Responsibly
	Bringing AI to Practice Management
	Making Decisions About AI Adoption