Members now engage by asking questions directly rather than conducting traditional searches.

Until recently, pension scheme members seeking information typically began their inquiries on the scheme website.

Members navigated menus, downloaded PDF documents, and consulted member guides in the expectation of eventually locating the required information.

If these efforts were unsuccessful, members would contact the scheme’s administration team.

This model functioned adequately for several years.

Currently, this approach does not align with contemporary information-seeking behaviours.

Research commissioned by Lloyds Banking Group indicates that over half of UK adults (approximately 28.8 million individuals) currently utilise artificial intelligence tools, such as ChatGPT, for financial matters. Money-related inquiries are the most prevalent use case for generative AI, with many individuals seeking information on pensions, investments, and retirement planning.

Member expectations have changed considerably.

A crucial consideration is whether pension schemes have adapted to these developing expectations.

The evolution of member support

The evolution of member support

Stage 1: Websites

The first generation of digital member engagement focused on publishing information.

Members were required to browse the following resources:

  • Scheme booklets
  • FAQs
  • Retirement guides
  • Forms
  • Policy documents

The information provided was generally accurate.

However, locating specific information was often challenging.

Stage 2: Frequently Asked Questions (FAQ's)

The subsequent development involved introducing FAQ pages.

These resources addressed common member inquiries, including:

  • When can I retire?
  • How do I update my address?
  • How do I nominate a beneficiary?

This represented progress over previous methods.

However, this approach was effective only when members posed questions that precisely matched the website’s anticipated queries.

Stage 3: Rule-Based Chatbots

The subsequent phase introduced chatbots.

To many organisations, these systems appeared innovative.

In practice, most chatbots guided users through predefined decision trees:

  • Press 1…
  • Choose A…
  • Did you mean…

The user experience was interactive, but rarely conversational in nature.

Members often experienced frustration when their inquiries fell outside the chatbot’s programmed responses.

Stage 4: Generative Artificial Intelligence

The introduction of ChatGPT signified a significant transformation.

For the first time, individuals could pose questions directly, such as:

  • "Can I retire early?"
  • or "Should I combine my pensions?"
  • or "Why is my pension worth less than I expected?"

Rather than navigating complex website structures, artificial intelligence systems could interpret natural language queries.

This development constitutes one of the most significant changes in digital customer experience in recent decades.

Members now increasingly expect all organisations to provide similar capabilities.

However, a significant challenge remains.

ChatGPT. Claude. Gemini. Copilot.

These platforms share a common limitation.

They have extensive general knowledge but lack specific information regarding individualpension schemes.

They don’t know:

  • Your trust deed
  • Your member handbook
  • Your retirement process
  • Protected pension ages
  • Discretionary policies
  • Internal escalation routes
  • Scheme-specific communications

Instead, these systems generate responses based on patterns identified in publicly available data.

Frequently, these responses are helpful.

However, in some cases, the information provided is inaccurate or incomplete.

Which? recently evaluated leading artificial intelligence systems on financial questions and identified numerous inaccuracies, including incorrect guidance regarding ISA regulations, taxation, and consumer rights. The Financial Conduct Authority (FCA) also reminded consumers that general-purpose AI tools are not regulated financial advisers and do not offer the same consumer protections.

Stage 5: Governed Artificial Intelligence

This stage represents the beginning of the next evolution in member support.

This approach does not involve the introduction of another chatbot, website, or FAQ page.

Instead, it launches a governed Pension Assistant.

MemberBridge integrates the conversational experience that members expect with the governance standards required by their individual pension scheme rules.

Rather than relying on general internet knowledge, responses are based on:

  • Your documentation
  • Your scheme rules
  • Your approved communications
  • Your governance framework
  • Your escalation policies

When appropriate, MemberBridge can:

  • Explain concepts in plain English
  • Provide links to supporting documents
  • Recognise vulnerable situations
  • Escalate to an administrator
  • Maintain a complete audit trail

The outcome is not limited to expedited responses.

It also fosters greater confidence among both members and pension schemes.

The Distinction Lies Beyond the Artificial Intelligence

A common misconception is that MemberBridge is in direct competition with ChatGPT.

This is not the case.

Large language models represent a significant technological advancement.

These models have fundamentally transformed information consumption patterns.

The primary innovation does not reside within the model itself.

Rather, it is found in the surrounding governance and contextual frameworks.

Recent academic research on trust in financial artificial intelligence has identified four primary drivers of confidence: human oversight, transparency, accuracy, and governance. Confidence is not achieved solely by making the AI appear more human.

This challenge is particularly relevant for pension schemes.

The Future of Member Engagement

Members are expected to continue utilising ChatGPT.

This trend is unlikely to change in the foreseeable future.

However, the central question is distinct.

When members seek information about a specific pension scheme, it is important to consider their preferred source of information:

  • Should responses originate from an artificial intelligence system trained on publicly available internet data,
  • Or from a system governed by the specific pension scheme’s policies and documentation?

The evolution of member engagement isn't about replacing AI.

Rather, it focuses on providing artificial intelligence with the context, governance, and accountability required by pension schemes.

Sources

  1. Lloyds Banking Group – Research shows approximately28.8 million UK adultsuse AI for financial matters.
  2. Which? – Testing of ChatGPT, Gemini, Copilot and other AI tools on financial questions.
  3. Zarifis & Cheng (2024), How to build trust in answers given by Generative AI for financial questions.
  4. Lakkaraju et al. (2023), Can LLMs be Good Financial Advisors?