The ESG investing sector continues to grow. Bloomberg famously projected that more than $50T would be committed to ESG-driven investments by 2025 — one-third of all global assets under management. More than 700 asset owners are now UN Principles of Responsible Investment signatories, and Sustainability accounting standards organizations like SASB continue to gain influence. Amid the disparate drivers of this ESG investing boom, from the preferences of young investors to increasing board oversight of ESG metrics, the trend line is clear.  

Despite the rapid growth of ESG investing, portfolio managers and ESG analysts struggle to access truly convincing data about company-specific ESG performance. While the major rating firms do provide ESG ratings, those ratings tend to be woefully inconsistent compared to their credit ratings, with a demonstrable lack of strong score correlation across ratings agencies. 

To fill this vacuum, many serious ESG investors are seeking to enable their own proprietary ESG research and sorting capabilities. AI will become a critical tool in this effort to improve the accuracy of ESG ratings, particularly because it is uniquely suited to analyzing massive amounts of unstructured data — for example, text and tables buried in reports, regulatory filings, and even news stories. AI will provide the most efficient and comprehensive method for evaluating a company’s true ESG performance. 

The Challenge of Accurately Evaluating ESG

 As more assets flow into ESG-focused investments, the complexities of ESG rating and ranking only become more apparent. 

To begin with, no two investors are alike. Each investor has a unique set of ESG concerns and priorities, which means they will differ in how they rate the importance of specific KPIs. They also understand that the relevant ESG factors vary greatly across industries, geographies, and value chains, and that a one-size-fits-all approach does not do justice to the diversity of the companies they are evaluating. They would much prefer a model that allows them to configure data and analysis to suit their own differentiated perspectives and investing strategies. 

Then there is the inherent complexity of the data itself. Compared to the traditional financial data used by investment analysts, ESG data is both more fragmented and more varied. Numerous potential KPIs, data sources, and data formats are in play, including structured, semi-structured, and completely unstructured data. Furthermore, none of this data is static. Analysts need more than an annual or even quarterly ESG rating; they need to process data regularly to achieve a complete real-time view of a firm’s ESG performance. 

How AI Solutions Will Transform ESG Investing 

To leverage the power of AI in service of ESG investing, financial firms will need solutions that allow them to select the datasets appropriate for their proprietary analysis, perform AI-enabled ESG analyses of target companies, and obtain a simple and reliable ESG score based on their customized weighting.   

Natural language processing (NLP) has been employed in similar use cases for years, but in isolation it is not ideal for analyzing diverse data formats (text, charts, tables, etc.). Emerging GenAI capabilities will be able to refine the accuracy of these legacy AI models; for example, by summarizing the sentiment analyses performed by NLP tools and feeding those summaries back into the ESG scoring matrix. With GenAI and NLP working in concert, firms will be able to capture and analyze ESG signals that have previously been impossible to quantify, such as news feeds, community message boards, and other publicly accessible but non-standard and unstructured communications. 

AI will make ESG rating much more continuous and automatable, enabling firms to monitor the impacts of global events, new climate data, and ESG-related controversies on their portfolio and its constituent firms. Furthermore, AI will allow firms to craft their own investment theses by leveraging standardized yet detailed taxonomies and materiality maps to refine their weighting of more than 2,000 quantitative ESG sub-metrics. 

Yes, AI does come with risks, such as hallucination. ESG investors can mitigate these risks by adopting clear AI governance models. The Wipro WeGA (Wipro Enterprise Generative AI) framework, for example, insists on clear transparency guardrails. In the context of ESG investing, transparency guardrails can create mechanisms for fact-checking the output of a large language model (LLM) by providing links to the documents that the model used for the data extraction and analysis, and instituting methods for monitoring the underlying logic that drives the model’s conclusions. 

ESG-minded investors might also be concerned about the carbon footprint implications of AI models. While this is a genuine concern, the models appropriate for this use case are relatively modest in terms of their compute needs, and may be more energy-efficient than standard enterprise applications. 

AI, Data, and the Future of Sustainable Finance

Portfolio managers and ESG analysts are likely to be the most immediate beneficiaries of AI-driven ESG research and ranking, but these approaches and solutions will extend well beyond evaluating publicly traded equities and bonds. AI will be instrumental in the context of ESG lending and ESG insurance, helping banks and insurers vet the ESG performance of borrowers and clients based on factors like DEI performance and sustainability factors, and preparing them to meet future obligations as signatories of ESG pledges. AI-based solutions will also enable companies to more accurately measure their own ESG impacts as they seek to attract ESG-focused investors and meet new regulatory requirements. Furthermore, these AI-driven tools will enable a wider range of institutional, venture capital, and private equity firms — whether or not they are exclusively focused on ESG investments — to better understand the ESG positioning of their portfolios. 

Climate change, regulatory shifts, stakeholder pressures, and an increasing number of values-driven investors have contributed to an extraordinary rise in ESG-driven funds and firms. To maintain this momentum, ESG-focused investors need data platforms and tools to extract insights that can amplify the power of their ESG investment strategies. AI, and particularly GenAI, has arrived at just the right time to give investors the tools they need to make better investment decisions and achieve portfolios that precisely match their intentions.  

About the Author

Vikram Maduskar
Head of AI Solutions for BFSI

Vikram brings more than 20 years of global experience in banking and capital markets to his role as the head of Wipro’s global AI solutions practice for BFSI. He has focused on AI and data analytics in leadership and consulting roles with buy-side and sell-side firms in North America and Europe, and now drives applied artificial intelligence for use cases across financial services, including as the chief product designer of Wipro’s ESG investing solutions accelerator.