The landscape of sustainability reporting is evolving, and a significant stride was made with the release of the draft set of European Sustainability Reporting Standards (ESRS) by the European Financial Reporting Advisory Group (EFRAG) a few months ago. As part of our commitment to staying at the forefront of industry changes, our team delved into the intricacies of the ESRS, examining one of the datapoint documents released by EFRAG. Our objective was to transform these datapoints into a machine-processable format within our lab environment, paving the way for a preliminary analysis that sheds light on the potential scope of efforts required for effective reporting management.
Datapoints Analysis
Recognizing that the draft is pending approval, our initial analysis serves as an approximation, offering insights into the extensive reporting landscape envisioned by the ESRS. Our examination unveiled over 700 narrative-type datapoints, accompanied by 300 metrics demanding calculations, derived from various reporting sources, represented in tables, or amalgamations of these components. Additionally, there are approximately 100 datapoints that fall between these two extremes or are of a random nature.
Machine Learning Integration
With our v0 model now machine-processable, the allure of employing artificial intelligence became irresistible. Leveraging OpenAI’s GPT 3.5-turbo and 4.0-turbo models, we embarked on a journey to generate example narratives aligning with datapoints and formulate calculation formulas for metrics. While the accuracy of these formulas may vary, given the models’ ability to derive them from datapoint titles alone, these “version zeros” provide early indications of essential data elements and sets crucial for the final reporting result.
Preliminary Report and Future Outlook
The culmination of our efforts resulted in a first version report encompassing all datapoints, albeit an unlikely scenario in real-life reporting. Impressively, this comprehensive report spanned an extensive 650 pages and could humorously be deemed the “CSRD reporting template.” Looking ahead, we anticipate leveraging AI models to deepen their understanding of reporting requirements, paving the way for predictions regarding the source systems of the data. This, in turn, will facilitate the creation of initial data models for corporate IT, offering a glimpse into the magnitude of the impending changes on the horizon.
As we eagerly await further developments, stay tuned for updates on our journey into the heart of the ESRS and the evolving landscape of sustainability reporting. The integration of AI promises to reshape how we approach reporting requirements, setting the stage for a more streamlined and efficient future.