Three quarters of pension trustees have stated that they use artificial intelligence (AI) within their work, Aon has found.

The professional services firm’s survey of 61 UK defined benefit scheme, ranging from under 1,000 members to over 20,000, asked about the adoption and usage of AI across governance and trustee duties, as well as their awareness of its use among advisers and suppliers.

Aon found that 76% of respondents reported some use by trustees, with 65% having discussed AI opportunities. However, just 18% have recorded potential opportunities.

No respondents reported a documented roadmap balancing the benefits, risks and controls of AI use, while 53% reported that trustees had received no AI training. Meanwhile, almost two thirds (61%) said they do not know how their advisers use AI.

Almost three quarters (73%) carry out no specific assessment of provider of AI governance, while 94% believe that their members are using AI to get answers to questions about their pension benefits.

Partner at Aon, Paul McGlone, said the survey provides a “timely snapshot of usage and understanding” among defined benefit schemes.

He added: "It makes it clear that pension schemes are already working with AI, whether through trustees, advisers or wider member behaviour. It’s no surprise that it appears to be the larger schemes that are leading the way.

"But it’s also apparent that AI’s use is relatively ad hoc and wider knowledge of its usage is limited. While trustees are already making use of it for their own purposes, there would appear to be questions over how much understanding and visibility they have of its use by the advisers and suppliers they work with and who provide services to their scheme.”

"We know that the Pensions Regulator is due to issue guidance on this area very soon and I’m sure that this will provide a constructive framework for schemes to follow. The priority should not be to stop useful innovation, but to make it visible, controlled and accountable. Trustees should start with the basics – they need to understand where AI touches the scheme, that their data is protected, that they have challenged providers on their usage and also made sure that it’s still a human who is responsible for decisions and member outcomes."

---