All pension firms use artificial intelligence, a new survey suggests, with improving efficiency and reducing cost seen as the greatest opportunity the technology offers, followed by closing the advice gap. As savers increasingly turn to AI for advice and guidance, a comparison of human and AI recommendations by Aon has yielded some concerning results. 

The Society of Pension Professionals’ AI Survey 2026 found that 100% of pension firms are now using AI, and more than two-fifths of 400 pension professionals attending a recent SPP event see the technology’s greatest opportunity for positive impact in pensions as improving operational efficiency and cost (42%). 

Over a third (34%) said AI can support member advice and guidance, helping to close the advice gap, while about one in 10 cited enhanced data quality and analysis (12%) or improving advice quality and reducing manual error and rework (11%). Just 1% sees the greatest opportunity as faster identification and resolution of complaints and breaches.  

SPP member Helen Howell, a partner at LCP and chair of the event, said: “The results of this industry polling highlight that pension professionals see AI not simply as a new technology, but as a practical tool for addressing some of the sector’s most pressing challenges.”  

Howell said it was particularly encouraging that so many pension professionals see AI’s potential to support member guidance and help close the advice gap. 

“As AI adoption becomes universal across the pensions industry, the focus is increasingly shifting from whether firms should use AI to how they can use it responsibly and effectively to deliver better outcomes for schemes, employers and members,” she said.

The role of AI in retirement decision making is increasingly in focus as people turn to LLMs to find information and guidance. Provider PensionBee said traffic to the government’s MoneyHelper website has dropped, while AI overviews referring to it have gone up. However, it said there has apparently been no drop in traffic to PensionWise, suggesting informational guidance is more vulnerable to AI than transactional services.
 
In a recent publication, consultancy Aon placed AI output on retirement planning and human advice side by side to compare them. 

“It has provided us with some fascinating insights into how members’ retirement decision-making may be changing as AI use continues to become more mainstream,” said partner Kelly Hurren.  

For example, the exercise found that the same information fed to the same AI tool produced different recommendations and reasoning on different occasions, which could lead to very different actions and outcomes. Aon also said the AI could easily be steered to support a preferred recommendation the custormer is set on with very little prompting, while a human adviser was seen to challenge a customer’s views much more robustly. 

There were also issues with calculations, which could be oversimplified or inaccurate. Factually incorrect claims were presented confidently and authoritatively, with complex justifications. Aon also highlighted the speed of the answer as a possible issue, potentially leading to rash decisions.   

“The results were mixed, but it’s clear that AI will become a significant part of how members navigate retirement. It has the potential to improve understanding and engagement – but it also introduces new risks. We believe that schemes and employers which recognise this early can shape and support strategies that blend the best of digital capability with human expertise,” said Hurren. 

Aon said as AI is not going away, pension schemes should equip members to use it safely and critically, communicating the risks in using it for retirement planning. They should also test how well AI summarises their online presence and materials, checking that summaries reflect what schemes want members to understand and the information the AI summary pulls out is up to date. 

The Investment and Saving Alliance recently warned of potential “widespread consumer harm” from unregulated artificial intelligence tools. The organisation wants the Financial Conduct Authority to introduce guardrails into AI tools and police the boundary between regulated and unregulated advice in AI – or alternatively, relax requirements on regulated advisers to level the playing field.