Project title
Business and the Nature Crisis: Analysing Corporate Discourse with Large Language Models
Research question
How do powerful corporations in the agrifood, construction, and mining sectors frame nature and biodiversity in their sustainability strategies?
Project description
Philip Schleifer’s project uses large language models (LLMs) to explore how major companies are responding to the global biodiversity crisis. Since the 1970s, wildlife populations have declined by nearly 70%, and large multinational corporations—particularly in the agrifood, mining, and construction sectors—have had an outsized impact on nature.
The project creates a new dataset and uses LLMs to investigate how some of the world’s most influential companies frame their environmental policies. For instance, it looks at whether these corporations take into account ecological, economic, socio-cultural, and intrinsic values in their biodiversity strategies.
By using LLMs as “research assistants”, Schleifer is tapping into a powerful new method for analysing large volumes of text—one that is quickly changing how social scientists work. Because this approach is still emerging, the project also tackles key challenges around research design (such as how prompts are written and how reliable the results are), ethical concerns (like bias and transparency), and legal issues (including privacy and data use).
Selected publications
- Schleifer, P., & Fransen, L. W. (2024). Smart mix politics: Business actors in the formulation of global supply chain regulation. Review of International Political Economy, 31(6), 1710-1734. https://doi.org/10.1080/09692290.2024.2367582
- Schleifer, P. (2023). Global Shifts: Business, Politics, and Deforestation in a Changing World Economy. ( Earth System Governance). The MIT Press. https://doi.org/10.7551/mitpress/14769.001.0001
- Schleifer, P., & Sun, Y. (2020). Reviewing the impact of sustainability certification on food security in developing countries. Global Food Security, 24, Article 100337. https://doi.org/10.1016/j.gfs.2019.100337