- Generative AI is changing how leaders harness unstructured text data—such as annual reports, contracts, customer feedback, technical documents, and internal memos—to generate actionable business strategy information.
- Previously, analyzing vast amounts of corporate text was time-consuming and costly, leaving much critical data “trapped” in long, difficult-to-exploit documents.
- Research fine-tuned a GPT model to analyze the “Business Description” section of 10-K filings from U.S. public companies to detect content related to products and services supporting decarbonization.
- The training dataset included approximately 3,500 labeled sentences from various industries to help the AI distinguish between selling climate solutions and merely mentioning environmental issues.
- The model was then applied to classify nearly 10 million sentences from 39,710 10-K reports of 4,483 U.S. companies between 2005 and 2022.
- Identified climate solutions include batteries, electric vehicles, energy storage, renewable energy (wind/solar), recycled materials, plant proteins, energy-saving solutions, and biofuels.
- Results show that the level of “climate solution intensity” extracted from the text correlates with higher revenue growth, greater market valuation, and positive stock market reactions to climate policies.
- Text analysis also helps detect early industry convergence, such as energy storage technology relating to both the automotive and electric utility industries.
- Industries with similar climate solution themes also tend to show similar stock fluctuations, indicating that business foundations are becoming interconnected.
- Generative AI also helps validate strategic assumptions, such as the political influence on the energy transition being lower than predicted as technology becomes cheaper and more efficient.
📌 In the past, analyzing large volumes of corporate text was time-consuming and expensive, causing much important data to be “trapped” in long and difficult-to-extract documents. Generative AI is changing how leaders exploit unstructured text data such as annual reports, contracts, customer feedback, technical documents, and internal memos to create actionable information in business strategy. Signals extracted from text not only reflect business activities but are also related to revenue growth, market valuation, and competitive trends. This method can be applied to many other areas such as supply chain, customers, human resources, and legal risks.
