RESEARCH · RESEARCH · #2846
MIT launches LLM Election Observatory to track chatbot responses about the 2026 midterms
Researchers at MIT, led by Professors Adam Berinsky, Chara Podimata, and Charles Stewart III, have built the LLM Election Observatory: a dashboard that runs a fixed set of election-related questions through popular AI chatbots every few days to document how their responses vary by time, location, and user attributes ahead of the 2026 U.S. midterms. The project aims to create a systematic historical record of chatbot outputs and to study subtler effects such as framing, emphasis, and how models shape voters' understanding of political issues.
KEY POINTS
- Researchers at MIT, led by Professors Adam Berinsky, Chara Podimata, and Charles Stewart III, have built the LLM Election Observatory: a dashboard that runs a fixed set of election-related questions through popular AI chatbots every few days to document how their responses vary by time, location, and user attributes ahead of the 2026 U.S.
- The project aims to create a systematic historical record of chatbot outputs and to study subtler effects such as framing, emphasis, and how models shape voters' understanding of political issues.
- Documenting how chatbots answer election questions over time matters because those responses can shape voter information, discourse, and potentially choices, and a systematic record enables research and policy responses.
WHY IT MATTERS
Documenting how chatbots answer election questions over time matters because those responses can shape voter information, discourse, and potentially choices, and a systematic record enables research and policy responses.