When public input is recorded but cannot change the plan
Researchers studied 22 documents from three public-sector AI projects. They argue that the way officials record public input can filter out objections and calls not to deploy. The study reads these records closely; it does not measure how often this happens. A useful test is whether dissent can change the decision.
What the source reports
Srijan Pandey and Devansh Saxena read 22 documents from three public AI projects. They asked how officials wrote down what people said. Their reading suggests that official records can soften or leave out strong objections. Some people wanted the AI plan stopped. The authors argue that a record of public input does not show who had power over the final choice. This is a close reading of three cases. It does not show how often other agencies ignore comments. The question for any one project is direct: did a call to stop reach the people in charge, and could it change their plan?
Original source
- Title
- Govern, Map, Measure, Absorb: The Legibility Trap in Public Sector Participatory AI Governance
- Author
- Srijan Pandey, Devansh Saxena
- Publication
- arXiv
- Date
- Tuesday, October 6, 2026