Generative AI can turn member dissatisfaction into a detailed complaint within minutes. Trustees should respond by strengthening the support provided before retirement decisions are made, writes Guiide’s Philip Hodges.
A recent report in The Times highlighted the growing use of artificial intelligence (AI) to prepare and strengthen employment tribunal claims. Submissions that may once have amounted to a short grievance can now run to multiple pages, citing legal references, multiple and nuanced allegations, and demands for redress.
Each argument needs to be considered and answered, often regardless of its accuracy or relevance, and it’s not too great a leap to suggest that pension trustees should take note.
A member does not need to understand pension legislation or trustee duties to ask an AI tool to prepare a plausible complaint. Using a simple prompt, they will, within seconds, be able to explore their scheme grievance and the associated personal impact.
Putting AI to the test
We tested this by asking Microsoft’s Copilot to draft a complaint from a member who had withdrawn their entire defined contribution pension as a lump sum and later experienced financial difficulty.
Within seconds, we had a formal complaint letter in hand, which we validated with a pensions lawyer. The adequacy of the scheme’s retirement communications was challenged, warnings were served about draining the fund, and questions were raised about guidance the member had received.
Scheme trustees were requested to review processes and consider redress, and were advised that a complaint was being raised under the scheme’s internal dispute resolution system.
The letter did not establish that the trustees did anything wrong and no wider context was considered. The member may have received all the information required and made personal choices as a result – but, of course, none of this was known to the AI agent.
It has quickly become much easier for members to put together detailed challenges, including arguments they may never previously have known to raise.
A growing risk to consider
Despite these potential gaps in understanding, once a complaint is raised, it must be investigated. The onus is on the scheme to retrieve the relevant communications and demonstrate the support that was made available. Done at scale, this quickly becomes time-consuming, resource-intensive, and expensive.
This risk will grow as schemes develop default retirement solutions. A default designed for the general membership or a broad cohort cannot know everything about one individual. It may not account for their state pension income, other pension pots, ISAs, savings, tax position, desired income or plans for later life.
AI will be ready to identify every question that a member who receives a poor outcome could ask.
However, technology also has a significant role to play by helping provide better, proactive support before a difficult-to-change decision is made. Solutions exist that can help members understand their options and the consequences of different choices without crossing into regulated financial advice. Where a member still needs or wants advice, they reach that point with a clearer picture of their needs.
This approach cannot remove every complaint, but it can help trustees demonstrate that meaningful, accessible support was available when it was needed.
If AI gives members better tools to challenge retirement decisions after the event, trustees need better tools to support those decisions from the outset.
Philip Hodges is a director at Guiide.
