Invisible Giants: Mapping Fossil Fuel Industry Representation on Spanish Wikipedia and Generative AI
🎥 Session recording: [https://youtu.be/IKDrJJMUDW8?t=24888](https://youtu.be/IKDrJJMUDW8?t=24888)
This presentation synthesizes two research projects on fossil fuel industries. The first is a [large-scale multilingual](https://meta.wikimedia.org/wiki/Research:Fossil_fuel_industries_and_climate_change_on_Wikipedia) analysis led examining mentions of climate change in fossil fuel articles across 4 UN languages + Portuguese. The second is my Master’s thesis, where I did an in-depth qualitative analysis of seven major fossil fuel companies with presence in South America, comparing their Spanish Wikipedia entries with responses from ChatGPT, Google AI Mode, and Google AI Overview.
Our findings reveal a structural disconnect: corporate articles often acknowledge climate responsibility, but articles on climate change don't link back to specific industries. On Spanish Wikipedia, content quality is low, with promotional tones and outdated data, and a heavy reliance on corporate websites. These practices create important knowledge gaps, which GenAI tries to solve with mixed results. This includes citing social media (including YouTube and Meta-owned posts), while under-citing traditional media. Wikipedia and GenAI barely cited research. In comparison, while Wikipedia articles tried to remain descriptive, GenAI answered with an “economic development vs. environmental protection” framing to the descriptive prompts. This poses interesting questions around reliability, sourcing practices, and neutral point of view.
Fossil fuel companies have a long-standing tradition of participating in climate disinformation campaigns, and therefore they are a good limit case to test assumptions on how they are being represented in the current information ecosystem. The presentation will share high-level findings and conduct prototyping community-led oversight and regulatory frameworks to ensure reliable knowledge about corporations.
