What Role Can Artificial Intelligence (AI) in Sustainable Finance Play in Driving Real-Economy Sustainability Improvements?
Author: Minxuan Zhang
The below article is one of a series of blogs based on a Policy Brief shortlisted as a finalist for the 2026 Chronos Sustainability Prizes at the LSE.
Sustainable finance and sustainable investment strategies have attracted unprecedented amounts of capital over the past decade. ESG-related information (e.g., ESG ratings, corporate climate disclosures) is now widely used. Artificial intelligence (AI) is increasingly being promoted as the next frontier of sustainable finance. Its promises of better data, stronger verification, and more advanced risk analysis are being promoted as ways to bridge the gap between financial markets and sustainability outcomes. Of course, this argument relies on the assumption that the key barrier to better sustainability outcomes is a lack of high-quality ESG-related information.
Where AI Helps
Data on emissions, environmental incidents, and supply-chain impacts are often incomplete, selectively reported, or difficult to audit. AI can help to address these issues. For example, satellite imagery, geospatial analysis, and natural language processing can all help investors to verify sustainability claims against external evidence rather than relying solely on corporate disclosures. Machine-learning models can also help by estimating difficult-to-observe indicators (e.g., Scope 3 emissions), by analysing large and heterogeneous datasets, by identifying emerging transition risks, and by generating forward-looking assessments.
The Risk of Scaling Existing Weaknesses
Despite its promise, many of the weaknesses already present in sustainable finance remain embedded in AI-generated datasets. Models trained on disclosure-heavy ESG datasets tend to reward reporting volume and quality rather than operational performance. Likewise, with AI systems drawing from similar data providers, sustainability frameworks, and historical market patterns, they tend to converge on the same ‘green winners’; again, these tend to be large firms with extensive disclosures and strong ESG profiles. Therefore, rather than correcting existing weaknesses, AI may amplify them.
Sustainable Finance Has a Transmission Structural Problem
Information is, of course, integral to the efficient operation of markets. However, we need to recognise that the transmission from financial markets to corporate behaviour change is often weak, even with more accurate sustainability performance observation. Real-economy impact arises primarily through three channels: firms’ cost of capital, access to liquidity, and investor influence over corporate practices.
AI can improve the precision of sustainability signals flowing through these channels, but it does not fundamentally strengthen the channels themselves.
Most analysis of these transmission channels concludes that they do not work as effectively as they could from a sustainability perspective. There are multiple reasons. One is that many large emitters are not strongly constrained by financing conditions. They often have diversified funding sources, strong cash flows, and access to alternative capital providers. Another is that divestment frequently reallocates ownership without materially affecting firms’ operations. Even stewardship outcomes ultimately depend on ownership structures, coordination, and the willingness of investors to exert influence, rather than on analytical precision alone.
In other words, AI may sharpen aspects of the information signal, but the signal is not necessarily the main driver of action. That is, the real question for sustainable finance is not whether AI can improve sustainability information, but whether sustainable finance possesses effective mechanisms for translating information into real-world change. Better data may be necessary, but it is not sufficient.
Closing Reflections
Ultimately, AI can strengthen information flows, but information alone does not create impact.
We need to ensure that improved information is connected to mechanisms capable of influencing corporate behaviour. This starts with the users of data. Investors (asset owners and asset managers) need to be much clearer about where and how their influence is most likely to affect corporate behaviour (i.e., they should focus on the outcomes and effectiveness of their stewardship activities) and whether the decisions being made by companies (e.g., changing production processes) are actually delivering real changes in their social and environmental performance.
Notes
Minxuan Zhang’s policy brief, ‘What Role Can Artificial Intelligence (AI) in Sustainable Finance Play in Driving Real-Economy Sustainability Improvements?’ was highly commended for the Chronos Sustainability Undergraduate Prize 2026. The policy brief can be found here.
Minxuan has completed the 2025/26 General Course at the LSE and is pursuing a BA in Economics at the University of Southern California. Her interests include sustainable business, development economics, and the role of business in driving social and environmental impact.
