RESEARCH PROPOSAL
Why this research, and why now
Financial institutions already rely on artificial intelligence for ESG and sustainable finance work, both directly and through third-party service providers. Adoption is racing ahead of understanding: the opportunities, the risks and the governance it calls for are still poorly mapped.
Because AI is a general-purpose technology, organisations are taking very different paths with it. Pooling real user experience across the industry is the quickest way to spot good practice and keep risks in check, rather than every organisation learning the same lessons on its own.
What the research will do
- Examine how AI is changing ESG analysis and investment decision-making.
- Assess the opportunities and challenges that come with adopting it.
- Capture the governance, safe-use and ethical practices already in place.
- Set a baseline and recommend high-value, safe and ethical ways forward for the industry.
Where AI is already being used
We see five broad uses of AI in ESG and responsible investment work today. The examples are drawn from the full proposal (PDF).
| Use | Examples in ESG work |
|---|---|
| Information summarisation and extractionGathers, condenses and organises information from structured or unstructured sources for analysts. |
|
| Insight generation and thematic analysisDetects patterns, trends or narratives across data to surface emerging ideas or assess qualitative factors. |
|
| Analytical challenge and bias testingTests assumptions or offers counterarguments to an analyst’s conclusions. |
|
| Process automation and monitoringAutomates or tracks repeatable processes and compliance checks. |
|
| ProductivityMore efficient everyday working, outside the investment process itself. |
|
The risks we need to manage
These risks are well recognised across the industry. How they play out in ESG work is far less well understood.
| Risk | Example in ESG work |
|---|---|
| HallucinationFalse or misleading outputs presented with confidence. | A sustainability report generated from made-up emissions figures or scientific papers that don’t exist. |
| Cognitive outsourcingOver-reliance on AI reasoning or outputs, dulling human scepticism. | An analyst relying solely on AI to assess a company’s ESG performance. |
| Inconsistency and basic errorsDifferent or contradictory outputs to the same or similar prompts. | AI miscalculating carbon emissions, or giving different ESG ratings for similar data sets. |
| Data and privacy exposureExposure or misuse of proprietary or personal information. | Staff uploading confidential company data and documents into a public AI tool. |
| Compliance breachesBreaches of regulation or firm policy, such as inadvertently giving advice. | AI-generated ESG reports that don’t meet reporting regulations or internal processes. |
What the industry gains
A shared, evidence-based picture of AI in ESG would:
- inform debate on the benefits and trade-offs of different uses
- show what best and emerging practice looks like
- support stronger governance and ethics standards across the sector
- give investors, service providers and regulators a clearer view of emerging needs, so everyone can move faster and more safely.
Method and timeline
We’re moving quickly. Right now we’re securing partners, refining the scope and reviewing the literature on AI adoption in ESG and related fields.
| When | What happens |
|---|---|
| September 2026 | Preparation and literature review. |
| October 2026 | A survey of finance professionals, plus focus groups and one-on-one interviews. |
| November 2026 | A report with recommendations for further work. |
| December 2026 | A living AI and sustainable finance research hub launches on Altiorem. |
Get involved
To make this research as useful as possible, we’re looking for people and organisations who can:
- Take part in a focus group or interview
- Knowledge partner (expert support and advice)
- Host a focus group or event
- Give access to investor networks
- Help with publishing and promotion
- Financial support
Prefer to talk it through first? Email Pablo Berrutti.
Banner image: “Sydney skyline at dusk” (December 2008) by Diliff, via Wikimedia Commons, licensed under CC BY-SA 3.0. Darkened by Altiorem for legibility; this adaptation is shared under the same licence.
