The DC decumulation landscape is changing rapidly, but not always in ways trustees and providers expect. While policy discussions continue to focus on improving pathways for savers, value for money and governance, savers themselves are taking matters into their own hands.
Faced with complex decisions at retirement, many are now turning to artificial intelligence tools to seek answers about how to access and use their pension savings. We know this anecdotally but also by talking to our friends and family – if you haven’t asked them how they get information about their pensions, then do! The response is enlightening.
This shift presents both opportunities and risks. AI-powered tools can make information feel accessible and personalised, but they also sit largely outside the territory of regulated advice and scheme oversight. As DC pots grow and decisions at retirement become more consequential, trustees and providers need to understand how savers’ behaviour is evolving – and how retirement propositions and governance must adapt in response.
Scheme communications, even when technically robust, can sometimes be difficult for savers to follow – especially if they are dense, generic or overly cautious. By contrast, AI tools offer something appealing: instant responses, plain English explanations and the sense of a tailored conversation. For a saver asking “how much can I safely take each year?” or “should I buy an annuity now?”, an AI chatbot may feel more approachable than a lengthy signposting document.
On a positive note, these aspects of AI support could help people to become more engaged with pensions. The risk, however, is clear. AI tools which sit outside the territory of regulated advice and scheme oversight may provide incomplete, outdated or inappropriate guidance.
The information they gather might not reflect a saver’s full circumstances and may blur the line between generic information and personal recommendation. The problem is that savers may not appreciate these limitations – particularly when answers are delivered confidently and conversationally.From a trustee and provider perspective, the growing use of AI by savers raises difficult questions. While schemes are not responsible for third-party tools, poor outcomes driven by misinformation could damage trust and confidence in the scheme itself and lead to complaints.
There is also a broader governance issue. Simply warning savers not to rely on third-party AI tools is unlikely to be effective; once a tool is perceived as helpful, it will continue to be used. So we expect trustees and providers will also have to focus on mitigating AI risks through the design of their official saver support tools and engagement strategy.
For trustees and providers who will accompany the saver through their long term decumulation journey, this call to action is clear. But for trustees who are not expecting to do so, the next steps may not be so obvious. How much of the engagement and design can and should they be doing themselves? At what point would it be better to signpost savers to third-party support? What level of due diligence should trustees undertake before making that decision?
The forthcoming requirement for trustees to provide guided retirement solutions to savers and the FCA’s new targeted support requirements will have a major impact on retirements in the future. At their core, these initiatives aim to help savers make better decumulation decisions – the very space where savers may otherwise turn to third-party AI tools.
It is widely hoped that guided retirement solutions, in combination with innovation in provider propositions, could help to reduce the overwhelming number of options that savers would otherwise have to navigate their way through.
Targeted support could also assist by providing savers with structured, decision-focused prompts at key stages, helping them understand their options without straying into personal recommendations.
If pension schemes can improve saver experience by means of such official support tools, it should mean there is less need for people to go looking for answers elsewhere. If so, these policies could be a game changer for huge numbers of people.