Key takeaways

The rapid integration of artificial intelligence across the investment and stewardship landscape has prompted Glass Lewis to release their report ‘The Foundation Matters More Than the Model’ on 10 June 2026 which advocates for a human-centric model that enhances human judgement with AI rather than replacing it.

• Enhancement, not replacement: Glass Lewis argues that while AI can vastly accelerate research and automate routine tasks, it must function as a tool to augment human expertise, as automated systems lack the strategic nuance needed for complex governance decisions.
• Competitive advantage lies in trust: Glass Lewis emphasises that AI cannot replace institutional trust and market standards. AI must rely on trusted data and be verified with human oversight.
• Framework for investor-grade data: Glass Lewis describes criteria for ‘investor-grade’ data to ensure reliable outputs.

Remuneration Committees must ensure that disclosures are verifiable as trusted data, and governance frameworks make AI outputs safe for institutional use, to prevent any automated processes used by investors missing key rationale.

Will proxy advisers be replaced by AI?

Guerdon Associates has previously noted that some institutional investors are relying on AI to supplant their proxy advice.

In response to this movement, global proxy adviser, Glass Lewis, outlined its outlook on AI in this recent release, explaining that AI alone is not enough. As we move away from software systems based on a per user model to generative and agentic AI, Glass Lewis opines that human-centric AI will define the next era of investment stewardship.

It argues that, while AI can accelerate research and automate certain workflows, it remains a tool to enhance expert judgement, not replace it. Given how dependent AI outputs are on source data, Glass Lewis places strong emphasis on data governance, keeping humans heavily involved in maintaining and validating data sources. It requires that data is “investor-grade” through a framework that aims to capture the efficiency benefits of AI while ensuring validity and accuracy of outputs.

Guerdon Associates have explored the topic of AI in a prior article evaluating its efficacy as a source of remuneration advice and, surprising as it may seem to some, largely agree with Glass Lewis’s perspective. Summarising our review, AI is a valuable tool when used to augment and enhance work, but it cannot yet replace an experienced professional. Findings derived from AI still require a verification step and human input to ensure that outputs are valid for the specific scenario, accurate or simply not made up.

Why does this matter?

Investment stewardship teams consider a large amount of data from different sources including corporate disclosures, board and committee structures, executive compensation metrics, historical voting outcomes and more. Complexity comes not just from the volume of data but the requirement to consistently interpret different sources of data.

If proxy research teams and portfolio managers rely on different underlying datasets, the speed of automated AI tools can amplify contradictions. This data mismatch can easily cause different arms of the same investment firm to reach conflicting conclusions.

This could mean that the right key developments or rationale are not identified from remuneration reports or other disclosures. To mitigate, Glass Lewis again reiterates the importance of ‘investor-grade’ data.

How do you make sure your data is ‘investor-grade’?

Glass Lewis describes investor-grade data through the following characteristics:

• Accurate (validated against sources)
• Consistent (normalised under clear rules)
• Complete (coverage gaps identified and managed)
• Traceable (linked to sources and has processing history)
• Valid (conforms to defined formats and logic checks)
• Timely (updated fast enough for decision cycles)

These characteristics are further underpinned by security (protected through access controls) and observability (monitored over time for exceptions, drift and anomalies).

Glass Lewis’ approach

Glass Lewis opt for a human-centric approach built on governed data. By employing this approach, Glass Lewis aims to capture the efficiency benefits of AI while without foregoing “… transparency, traceability, or trust”.

This means involvement and accountability from data stewards, analysts, methodologists and governance professionals on all AI-powered workflows and processes.

Both AI and proxy advisers suffer from zeitgeist; it wins over logic

The prompts that investors like JP Morgan use in their AI agent replacing proxy advice are probably as flawed as proxy advisers’ guidelines.  They suffer from beliefs of what good governance looks like, not what actually delivers sustainable shareholder returns. Practically, we know that it is most efficient for proxy advisers to have their analysts strictly follow a set of rules. The rules are based on beliefs on what constitutes good governance. Great for consistency and efficiency, not for judgement. At some time or another almost all board remuneration committees as well as investor portfolio managers come to realise this.

© Guerdon Associates 2026 Back to all articles