As artificial intelligence influences more day-to-day operations across investment organizations, public pension fund sponsors report being cautious about its use for investment functions and plan administration, according to a recent National Conference on Public Employee Retirement Systems study.

The NCPERS research brief, “AI Adoption and Strategy in Public Pension Administration,” found that 58% of DB plan sponsor respondents reported being “optimistic or very optimistic” about AI’s impact on DB administration over the next decade. Some 63% of respondents cited a “lack of in-house technical expertise to evaluate or oversee AI tools” as their the top barrier to AI expansion. Among all respondents, 43% said their top risk concern is cybersecurity vulnerabilities from AI systems or vendors.

“DB plan administrators understand the stakes [of] hasty integration of AI into [an] important operation,” says Matt Eckel, NCPERS’ director of research. “We see that reflected in the deliberate pace [by] which AI is being integrated.”

Administrative tasks and automation are the areas in which the highest portion of sponsors have most adopted AI, are considering adopting AI within two years, or are piloting or testing AI (84%), followed by member communication and customer service (81%). AI was most actively used in member services (16%) and most actively being considered in fraud detection and cybersecurity (67%).

Eckel says he is not surprised that AI has been integrated primarily into member communications and back-office administrative functions, as the technology is advanced enough and the stakes are low, at least in comparison to using AI for investment decisions.

Sponsors Know Best

Respondents’ near-unanimous (96%) response that human judgment is the “primary driver of decisions where AI is used” underscores their emphasis on fiduciary responsibility and governance as they explore AI, Eckel says. He adds that public pension fund teams “seem to know what they don’t know and seem to be aware of the need to develop further internal capacity in order to fully integrate AI into their systems and operations.”

However, while very few public pension fund staff members can claim expertise in AI just yet, Eckel says many organizations have taken tangible steps to train their staff on AI. Signs of adoption and increased comfort with AI trickled in earlier this year.

The Big Picture

Hank Kim, NCPERS’ CEO, says the impetus for the new AI research came from a finding in NCPERS’ 2026Public Retirement Systems Study, released earlier this year, that showed a dramatic increase in AI adoption among public pension funds in 2026.

Respondents to the PRS study reported using AI most often in fraud detection and prevention (28%, up 25 percentage points from last year’s study); predictive analysis for actuarial forecasting (26.8%, up nearly 25 percentage points); participant communication and customer service (26%, up 14 percentage points); enhanced data modeling for investment opportunities (26.6%, up almost 24 percentage points); and automation of administrative tasks (25.8%, up nearly 15 percentage points). For each use case, between 58% and 66% of respondents reported that they were considering using AI applications in the future.

Kim says cautious optimism among public DB plan sponsors is “not to be mistaken for a lack of enthusiasm,” however. Deliberate pacing is “part and parcel of the character and culture of public plans,” and the public DB sector might just need to “test the waters” first.

Stephen Rosenberg, a partner in boutique law firm Rosenberg & Fiore LLC, says many service providers have created technology but are in the process of “fine-tuning it” before it is deployed. Pension funds, unless they have a trusted vendor, are not necessarily in a rush to “accept the black box result and feel comfortable that it’s a safe call,” according to Rosenberg.

Read more @cio