Professor Irena Vodenska, the Boston University Metropolitan College chair of the Department of Administrative Sciences and director of Finance programs, recently joined Bloomberg Intelligence’s ESG Currents podcast to lend her specialized expertise to a conversation surrounding the environmental, social, and governance, or “ESG,” elements of business operations, and the ways artificial intelligence can augment current capabilities to achieve greater policy results.
Along with several colleagues, Professor Vodenska recently completed a study into the way artificial intelligence can be leveraged to help navigate the rules, standards, and guidelines that encompass ESG. Joining as collaborators on the study were Associate Professor Lou Chitkushev, of BU MET’s Department of Computer Science, Visiting Research Professor Dimitar Trajanov, originally of the Saints Cyril and Methodius University in Skopje, North Macedonia, Mr. Risto Trajanov, and Mr. Georgi Lazarev. The study was published in Sustainable Investing: Problems and Solutions, a 2024 publication edited by Anatoly B. Schmidt, in a chapter titled: “Navigating global ESG investment regulations using AI.”
It’s a difficult field to manage, Dr. Vodenska explained, rife with what she called a global “alphabet soup” of misaligned regulations, some of which are voluntary, and an ever-changing legal landscape. She cited environmental regulations, in which some countries mandate disclosures, and other countries do not make disclosures mandatory but rather voluntary.
Professor Vodenska explained to the audience of ESG Currents, a Bloomberg Intelligence program aimed at demystifying the increasingly important realm of ESG, that her findings suggest the keys to effective regulation are the alignment of one or more of the 26 topics laid out in the SASB ESG Standards, the degree of enforcement, and whether the regulation promoted compliance or reporting. As an example of effectiveness, she cited the EU’s Corporate Sustainability Due Diligence Directive. Globally, there are plenty of ESG regulatory documents, and in the study Dr. Vodenska and her team extracted a subset of ESG regulations using the ESG regulations and reporting standards tracker, then analyzed over 100 relevant global regulations.
SASB, Dr. Vodenska explained, lays out five major themes categorizing existing regulations: environmental, social capital, human capital, business model and innovation, and leadership and governance. Most dominant topics were greenhouse gas emissions, energy management, and ecological impact. Less commonly addressed were issues like customer welfare and competitive behavior. They found that the European Union leads in effective and mandatory regulation coverage, addressing 20 of 26 SASB topics, with Saudi Arabia and Nigeria next in line with 19 topics addressed each. While these nations were most aligned with ESG regulatory themes in the mandatory disclosure regulatory space, for non-mandatory but effective regulations, the US ranked highest, with 21 topics addressed, with Malaysia and Mexico trailing close behind, with 20 each.
Asked what countries most surprised her in her research, Dr. Vodenska singled out the Netherlands and Switzerland. “They were notable outliers because despite their reputation for progressive policy, they had very low coverage in global, non-mandatory, ESG regulatory documents.”
As part of her research, Dr. Vodenska used both keyword-search based processors and large-language-model AI. The AI was able to accomplish much more than its keyword counterpart, thanks to the vital use of “RAG,” or retrieval augmented generation, which leverages real-time external knowledge sources to validate and vet findings.
Perhaps the most profound aspect of Dr. Vodenska’s findings was the potential for transformation posed by the arrival of AI, which can drastically reduce the amount of hours required to accomplish important ESG work. “[This work] is so much more difficult to imagine being done by just a human in real-time. Because it does take time to sift through the documents, to read the documents, to understand them. To understand the nuances,” she said. “The ChatGPT-based models outperformed the keyword searches because they understand context and nuance, not just specific words. They can detect implicit references to ESG topics, even when terminology may vary and it’s differently presented.”
Describing why RAG was such an important tool, Dr. Vodenska said, “It’s because it is able to retrieve up-to-date information, unlike static models that can only know what they were trained on, RAG . . . allowed real-time integration of current facts.”
“[RAG] can incorporate specialized content, like laws, financial disclosures, that general models might not. And it does reduce hallucinations—by grounding the answers in real documents,” Dr. Vodenska said. “The retrieval augmented generation helps minimize the risk of the model making things up.”
After all, for all its considerable prowess, the artificial intelligence has its limitations.
“The AI is here. And it isn’t meant to replace our critical thinking or the judgement or the expertise of the human ESG analyst,” she said. “We’re not advocating that the human is replaceable. Instead, we’re advocating that the AI is intended here to be really powerful tool, and to help the ESG analysts.”
With this robust tool now available, Professor Vodenska sees great potential for the future. “[We are] advocating for standardization, helping corporations report better—not necessarily more or less—and standardizing the ESG ratings across the board.”
Finally, Professor Vodenska spoke of her next research topic: investigating the detection of “greenwashing,” by corporations.
Visit Bloomberg Intelligence’s ESG Currents for more.
Listen to Professor Vodenska’s podcast appearance on:
Apple: https://lnkd.in/edJXAs26
Spotify: https://lnkd.in/eEaktrAK
Amazon: https://lnkd.in/eG-VgTVe
Bloomberg Terminal: https://lnkd.in/eynAh4Xs