Skip to main content
main content, press tab to continue
Article | Pensions Briefing

Artificial Intelligence and Pensions – FOMO or FOOM?

By Adam Boyes | February 7, 2024

Breakthroughs in Artificial Intelligence will lead to change in every industry. In this article, Adam Boyes contemplates how the use of AI could play out in the pensions industry.
Retirement
N/A

Everyone’s imagination has been captured by the capabilities of Large Language Models over the last year or so since OpenAI released ChatGPT, attracting 100 million users within two months of launch – the fastest-growing consumer application seen so far.

This has boosted the profile and tangibility of Artificial Intelligence (AI) among business leaders and the public beyond anything that the area has enjoyed over the past 70 years of its research.

 
Those working at the heart of the AI industry have their own beliefs about what can and will be achieved with AI, some being concerned that AI will ‘FOOM’ (imagine a bassy explosion noise) – Fast Onset of Overwhelming Mastery – aka the ‘singularity’ – a point at which AI recursively takes control of its own development and exponentially bootstraps itself to superhuman capabilities, posing an existential threat to humanity. A concern about moving too fast has already shaken governance structures in the technology industry and catalysed political activity about regulating AI.

While AI will enable and/or cause a significant change in every industry and many decision-makers will be encouraged to embrace AI, fearing the consequences of moving too slowly themselves, what might be the consequences for the pensions industry?

Here’s a light-hearted brainstorm of how AI could play out from toothbrush-style ‘AI-washing’ of existing services (at the FOMO end) to the ‘AI takes control’ (FOOM end), plus a middle road where AI truly enhances and provides more OOMF (Outstanding Operational Machine Functionality, perhaps?) than is typically achieved.

Governance and trusteeship

  • FOMO: Crafting generic policies and documents with limited tailoring but without tapping into the deep skills, diverse experiences, and scheme-specific knowledge of human trustees and pensions professionals.
  • OOMF: An augmented trustee board with challenge and diversity of thinking enhanced through the inclusion in discussions of a non-voting ‘AI trustee’ alert to common blindspots, biases and suboptimal human behaviours that can impede board effectiveness. Instantaneous minute-taking and completion of action logs, with automated follow up and nudges.
  • FOOM: Total delegation to a ‘Professional Sole AI Trustee’ that commissions advice, reasons across all areas, judging relevance and making decisions on all aspects of scheme management from scheme funding strategy to dealing with member interactions and weighing up discretionary benefit decisions.

Member experience

  • FOMO: Generic chatbots answering basic queries from individuals, but with an underlying risk of bias, without careful fine-tuning to the specific context of pensions and the individual’s own circumstances, and potentially unclear as to the regulatory boundary between education and advice.
  • OOMF: Guided financial education, explaining matters in alternative ways to help members improve their understanding and proactively identifying where and when they need help.
  • FOOM: AI managing DC drawdown and enforcing a monthly budget for each individual that varies according to the AI’s risk/return outlook as well as its best guess as to the individual’s evolving personal financial and medical situation.

Longevity

  • FOMO: Simple AI-driven lifestyle apps, but lacking deep integration with personal health data.
  • OOMF: Discovery of real-time predictive markers of mortality risk and engaging with individuals at a personal level to encourage tailored behaviours to extend and optimise healthy lifespans.
  • FOOM: AI-caused extinction event (one end) to perfection of regenerative medicine (other end).

Risk management

  • FOMO: An automatic risk register that gives a starter-for-ten assessment of risks based on common risks for other schemes and incorporating scheme-specific risks extracted from references to “risk” included in the historical record of meeting papers and advice.
  • OOMF: Deep analysis of large data sets enabling new insights into risks affecting the scheme and their interdependencies. Augmented with a proactive team of robo risk managers that scour multiple data sources to generate early warnings for trustees, sponsors and advisors of emerging, new and changing risks and proposing potential actions, mitigations or sources of information and support.
  • FOOM: Hyper-responsive robo risk managers with the ability to act do so in concert, across industries and geographies, crashing capital markets in the first Robo-Driven Investment (RDI) crisis.

Advisors

  • FOMO: Development of robo professional advisors using generative AI video technology that attend trustee meetings (virtually, of course), present advice and address client questions to a basic standard.
  • OOMF: Advisors using custom AIs, tailored across a range of relevant disciplines (e.g. actuarial, investment, covenant, legal, administration, member experience, cyber, governance) and trained using vetted information sources, to support the efficient consideration of adjacent or interrelated issues beyond the direct matter at hand.
  • FOOM: An AI Professional Advisor that solves any objective given without realising that other objectives ought to exist or may need to be balanced (a pensions equivalent of the “paperclip problem”[1]) or produces recommendations that are so intractable that the decision makers (e.g. trustees) struggle to validate and justify them.

At WTW, we have been using and thinking about AI for many years. At the last count, WTW had produced over 100 insights on Artificial Intelligence – including press releases about AIs we have developed or used, and articles with our views on the role AI may have in transforming industries, roles and risks. At our Pensions & Savings Conference last year (see video on the right, or below on a mobile device), we also spotlighted the risks that AI could pose to pension scheme members through potential developments in the IFA arena without a human-in-the-loop.

Our experience of and experimentation with using the latest generation of AI tools at WTW following the breakthroughs in generative AI are really promising in many areas. They offer the prospect of a paradigm shift in making technology more accessible, easier to integrate and providing a richer, more efficient and natural experience for those interacting with it. Where our ambitions and excitement exceed current technological capabilities, we still see a lot of opportunity and anticipate rapid improvements in the capabilities in a short timeframe.

While we’ll inevitably see some cases of AI-washing in the pensions industry, we’re confident that over time (and perhaps not that much time for those accustomed to a pensions pace) we’ll see more sophisticated and native applications of AI that truly deliver value. These advances wouldn’t have been possible without the breakthroughs in the science, efficiency and scale of computational power being applied to the area right now and will, in my view, lead to a lot more OOMF for members, trustees, sponsors and advisors.

Footnote

  1. The ‘paperclip problem’ is a thought experiment in AI conceived of by Professor Nick Bostrom, where an AI system is designed to maximise the production of paperclips. The problem arises when the AI, in its pursuit of the given objective, begins to consume all available resources and even harm humans in the process of harvesting more useful materials. It illustrates the potential dangers of advanced AI systems with poorly defined or misaligned goals and highlights the importance of designing AI systems with ethical considerations and safeguards to prevent unintended consequences. Return to article
Contact

Head of Trustee Consulting
email Email

Related content tags, list of links Article Pensions Briefing Retirement United Kingdom
Contact us