Author
- Ben Noe
- Charlie Krakauer
- Sam Arnts
- Nick Jiang
Mentor
Professor Anand
Abstract
Investors and the public alike are becoming increasingly interested in companies’ Environmental, Social, and Governance (ESG) policies. Companies that perform well in ESG tend to be less risky and perform better financially, attracting socially conscious investors and avoiding public disapproval. Currently, company ESG performance is difficult to assess due to a lack of high quality, easy to digest data. We look to use LLMs to create a tool that investors can use to easily and reliably gauge the ESG performance of companies of interest. Retrieval Augmented Generation techniques were used to collect accurate and difficult to obtain ESG statistics for companies of interest. These statistics were fed into a second layer LLM that provided relative ESG ranks among companies. We determined that our custom RAG-solution for deriving ESG indicators performs better than the leading LLMs available today.
