[SPONSORED CONTENT]
Now people can ask a machine about retirement, money and the choices that shape their future, is there a risk that pension providers get pushed to the side? This is one of the questions raised at the Mercer Master Trust’s (MMT) recent AI event ‘The Power of a Confident Voice’.
It is a fair question. A tool that gives an immediate answer can feel easier to use than a traditional provider channel. From the delegate discussions, it became clear that the appeal is not just speed. It is also the sense that the tool is ready to listen, without judgement, at any time. As one delegate explained:
____________________________________________________________________
“AI or a conversation with a tech agent allows you to ask anything you want without judgement.”
____________________________________________________________________
Pensions can be complicated, and for some people, talking to an adviser or provider can feel intimidating. Members may worry about saying the wrong thing or being judged for not knowing enough. In that context, a conversational tool can feel more approachable than a formal route to guidance. Add in the cost and effort involved in seeking help through more traditional channels, and it is easy to see why AI is starting to fill a gap.
What this means for the industry
This is where the risk of a growing distance between a member and their pension scheme starts to become serious. If a member can get a quick, confident sounding response elsewhere, what is left for the provider?
The answer from the webinar was thankfully not ‘nothing’ but it was certainly not ‘complacency’ either.
Providers cannot assume members will automatically come to them with their pension questions. If the customer experience is confusing, slow or hard to navigate, people will look for easier ways to get an answer. In that sense, the risk isn’t that the provider disappears overnight; it’s that it becomes irrelevant at the point when the member most needs help.
The risk to members
While recognising the opportunity AI presents, delegates were also quick to note the risks for members.
A polished answer can sound more certain than it should, which makes it easy to accept without challenge. The event repeatedly returned to one idea: confidence can be persuasive, but persuasion is not the same as accuracy.
The risks identified included incorrect answers, crossing the advice boundary, poor treatment of vulnerable customers, privacy concerns and weak accountability. One of the strongest points made was that the danger is not only technical error; it is silent harm – a member accepts the answer, never checks it, and only finds out later that the guidance was wrong or misleading. As one delegate noted:
____________________________________________________________________
“If that’s their only source of information and they’re not fact checking it, there’s a risk that they misinterpret it as well as it not being accurate.”
____________________________________________________________________
Members may use conversational tools to get started, but that does not remove the need for credible pension expertise. If anything, it raises the bar. Providers will need to be clearer, faster and more helpful, while staying firmly inside the boundaries of good governance.
Overcoming the risks
The good news is that discussion at the event was not just about risk. It was also about how to manage that risk and how to build AI that people can trust when discussing their retirement options.
The strongest theme was that confidence has to be earned through design, evidence, and human oversight, not just a polished tone of voice. Additionally:
- Use clear guardrails so the AI stays inside a defined pensions scope and does not drift into advice.
- Ground the tool in the right sources and make it obvious to members that it is scheme-specific rather than a generic chatbot.
- Design the AI to ask questions as well as answer them, so it can draw out context and help members get to the real issue.
- Build in testing, real-time monitoring, and escalation routes so the system can flag risk, hand off to a person, or pause when needed.
- Explain privacy and data use in plain English so the technology feels practical, helpful, and safe.
- Show visible human oversight with named accountability, so members know there is a real owner behind the system.
As Matt Gosden, Head of AI, MMT, explains: “The feedback from the event was clear. The industry will not win by shouting louder than the tools members are already using. It will win by being explicit about privacy, data use and accountability, so members understand who is responsible and what happens if the system fails.”
This is the real challenge. Not to out talk the technology but to make sure the provider remains the trusted source when the stakes are high.
The real risk is being bypassed
So, will AI disintermediate pension providers? The short answer is: not completely, but it can bypass them. The real threat is not total replacement. It is being skipped over by a better experience. That is why the industry needs to design for trust, keep humans in the loop and make sure its expertise is available when members need it most. This is one of the central themes of Mercer’s paper ‘AI consumer expectations and pensions’ the providers that stay relevant will be the ones that combine confidence with accuracy, and convenience with trust.
To read more articles from Mercer, visit the content hub on Corporate Adviser – here.