MIT Launches LLM Election Observatory Dashboard, Tracks 12 Models Across 19,000 Political Prompts
Summary
MIT unveiled the LLM Election Observatory on Sept 10, a public dashboard tracking how nearly a dozen major LLMs answer midterm political queries. Led by MIT assistant professor Chara Podimata with political scientists Adam Berinsky and Charles Stewart III, the team runs automated sweeps of ~19,000 prompt-identity combinations (varying stated gender, race, location, political lean). Early finding: ~30% of English queries across six chatbots return incomplete, unclear, inaccurate, or outdated answers.
Shared on Bluesky by 2 AI experts
-
Justin Hendrix @justinhendrix.bsky.social
"For nearly a month so far, the researchers have regularly run automated sweeps of all the possible combinations of prompts and identities — 19,000 separate queries each time. They designed a public dashboard called the …
View on Bluesky →
Originally reported by nytimes.com
Read the original article →Original headline: MIT Launches LLM Election Observatory Dashboard, Tracks 12 Models Across 19,000 Political Prompts