Bringing clarity to ESG data
Environmental Finance: Please introduce Clarity AI and explain how you generally apply AI.
Ángel Agudo: Clarity AI is a global sustainability tech company that uses technology and AI to provide data-driven environmental and social insights for decision-making to financial institutions, corporates, governments and end-consumers. Our clients in the investment segment manage over $60 trillion in assets, and include some of the world’s largest asset managers, like BlackRock. We provide value across three dimensions: data collection; methodologies to fill in the gaps and transform the data into useful information; and data management and analytics tools.
As a digital-native company, we apply technology and AI across our offerings, which include: scalable data solutions for extraction, mapping, quality assurance and aggregation; tailored insights and analytics for any sustainability-related use case; and seamless integration into clients’ workflows via widgets, API and standalone apps.
EF: What are the key shortcomings of ESG data?
AA: Clearly, there is room for improvement. According to a recent survey by Bloomberg, 63% of financial market participants consider quality and coverage issues with company-reported data to be their biggest concern when dealing with ESG data.
There are three key challenges that can only be solved through advanced technology. First, a lot of meaningful data is communicated in unstructured formats. Second, there is a lack of standardisation, with more than 860 ESG regulations and standards in place, according to the Principles for Responsible Investment. Not only that: companies may not report metrics consistently or they might present them in different ways. In fact, we found that one in five companies disclose different emissions data to CDP compared to their sustainability reports. Third, the poor quality of the data per se is also an issue, with potential errors from companies when doing their calculations and users when extracting data. Our research on data discrepancies revealed that 80% of errors found in data providers were due to manual work.
EF. How can AI help – and what are its limits?
AA: AI can truly be a game-changer in collecting information and analysing it in meaningful ways to better understand companies’ sustainability performance.
As an example, we use AI to increase data collection by 40% compared to the alternatives. It also allows us to validate data by comparing it with historical performance from the same company and its peers, and to improve our ESG data estimates to fill in the gaps where companies might not be reporting. We use AI to capture additional data: for example, our AI model processes more than 300,000 articles every day, classifying potential corporate controversies, connecting them with companies affected, and assessing their materiality. Most recently, we have launched new analytical and reporting capabilities to help our clients make sense of complex data and reduce time spent in reporting from hours of manual work to literally two clicks.
There are some limitations to AI when models are not properly trained and maintained, which can affect the accuracy of the results, leading to a loss of confidence. At Clarity AI, we always provide references to the source of the information and use retrieval- augmented generation to connect the results of our analysis with underlying facts.
Another challenge is the lack of expertise in AI models: they are typically trained using generic information. To ensure the accuracy of the results, which is paramount for decision-making and reporting, we mitigate this limitation by fine-tuning the models with domain experts. We have been training these models for over seven years, since our creation. For us, AI is not a recent development, it’s at the core of what we do.
A third challenge relates to the determinism and consistency of the results. These models typically are non-deterministic, which means that when they attempt to be creative with their results, they may introduce randomness. We prevent this by building guardrails that ensure our system doesn’t go in unintended directions.
EF: How should reporting companies respond to the greater use of AI?
AA: Companies are on a learning curve on ESG disclosures due to growing regulations and reporting frameworks. By 2028, more than 50,000 companies will fall under the Corporate Sustainability Reporting Directive, increasing internal complexities in ESG data collection and compliance processes.
We believe that companies should embrace AI to navigate this reporting landscape and the added pressure from other stakeholders. As experts in the field, we can enhance the precision and depth of companies’ ESG data management. We apply rigorous scrutiny to ensure data accuracy, connect data points to reveal valuable insights, and conduct assessments to fully understand the data’s implications. By providing this detailed context, we help them identify potential issues to refine their systems and improve their overall ESG reporting and compliance processes.
All in all, AI can support companies in achieving greater control and accuracy over their data across the value chain.
Ángel Agudo is Chief Product Officer at Clarity AI. For more information, see: https://clarity.ai
