While artificial intelligence represents a new cornerstone for institutional investors’ operational tasks, corporate governance will be essential to ensure the risks are properly managed, according to Andrew Roth, deputy director of the Teacher Retirement System of Texas.

During a webinar hosted last week by the National Institute on Retirement Security, he detailed how the pension fund already incorporated a governance structure related to the use of AI along its chain of operations. Through its AI guidance, the TRS recognizes both machine learning and generative AI tools and implemented a review process to approve tools and systems using these novel technologies. Moving forward, it has tasked a newly created multi-disciplinary review team with auditing any proposal involving data elements and the use of AI.

Read: AIMCo enters partnership to build AI, machine learning applications for investment management

“What they do is take in a lot of publicly available information, synthesize it, query and use that analysis as a result of all of that ingestion on information to help develop the building blocks for different strategic decisions related to investment opportunities, asset allocation, portfolio construction, a lot of different things that happen in the investment part of the organization,” he said, referencing the ways the investment team has already started implementing some machine learning AI.

After conducting a risk analysis on the TRS’ use of AI, the organization found it was most concerned by the unauthorized use or disclosure of sensitive or confidential information. In response to the potential risk, it has created mitigation strategies for each key concern surrounding the use of AI.

“The tools that have AI components built into them [have] great promise for transformational technology to quickly get things done and do things faster with fewer resources,” said Roth. “However, underlying that promise is a lot of risk.”

In order to evaluate how to approach the implementation of AI, the team first had to acknowledge it was already behind since the technology was already in its system through search engines and third-party software products and solutions. It put together use cases for AI in the typical model of a pension plan through benefits, investments and health.

Read: OMERS joining think tank to find uses for AI in the pension sector

Roth outlined how AI can already help a pension fund, with services like customer call support, which can become overwhelmed during busy periods such as tax time or retirement season. AI is also used to help oversee health plans managed by the TRS, he said, noting the technology creates predictive analysis to help establish premiums and respond to health trends quicker.

The risk associated with using AI can also show up through pension funds’ partners, depending on how they use the tools. “With all those different layers of vendors and software, having a really clear kind of plan and consistent approach to identifying those risks is something that’s really helpful and something that we paid a lot of attention to.”

Also speaking during the webinar, Nate Haws, associate principal consultant at Linea Solutions, said the improper use of AI can lead to increased cybersecurity concerns for pension plans. Any sensitive data used with AI assistant tools like ChatGPT can be used for future iterations of these tools, he noted, unless safeguards are put in place.

Read: Phishing and ransomware risks on the rise for pension funds, says expert