techpolicy.press web signal

Eric Gordon: AI won't rebuild public trust without listening

TL;DR

  • Boston University professor Eric Gordon argues chatbots can expand access to government services but cannot repair trust while agencies ignore public input they already collect.
  • He cites Partnership for Public Service data showing only 27% of Americans trust the federal government, and Pew finding 63% say AI is advancing too quickly.
  • NYC's MyCity chatbot, which gave landlords and employers illegal guidance before being taken down, is his cautionary case against automation-first strategies.

A Boston University professor argues in Tech Policy Press that AI tools like chatbots cannot rebuild government trust while agencies keep gathering public input and failing to act on it.

Eric Gordon, who directs the Center for Media Innovation & Social Impact at Boston University, writes that "The relationship between citizens and their public institutions is in crisis." He cites the Partnership for Public Service finding that only 27% of Americans trust the federal government, and Pew Research data showing 63% believe AI is advancing too quickly.

His cautionary case is New York City's MyCity chatbot, launched in 2023 and since taken down. "It gave answers to users that were wrong and in some cases illegal," Gordon writes, including telling landlords they could refuse housing vouchers and employers they could take workers' tips.

The counter-examples are offline. In Evanston, Illinois, after public health data revealed a 16-year life expectancy gap in the Fifth Ward, the city ran "data walks" where "people in the neighborhood had the opportunity to scrutinize the data, add local nuance." In Boston's Dorchester, community organizers built "On the Porch," curating trusted data sources from neighborhood leaders rather than relying on government chatbots.

Gordon's thesis sits in a single line: "people don't trust information; they trust messengers."

Shared on Bluesky by 2 AI experts