Voters think politicians don’t listen. Could AI help?
The student who answers the door looks a little hungover, and she may well be. It’s late August at East Stroudsburg University, a midsize public college in northeast Pennsylvania, and students are soaking in the final days of summer. But Kim Allen is undeterred by the girl’s half-opened eyes. A 64-year-old middle school Spanish teacher from New York City, Allen is particularly immune to the…
A 64-year-old middle school Spanish teacher from New York City, Kim Allen, is volunteering for Ground Truth, a project aiming to engage voters from all political backgrounds in open conversations. Primarily targeting the 2024 House seat in Pennsylvania, Allen listens to potential voters without pushing her agenda, aiming to understand their concerns and preferences.
She engages in conversations for over eight minutes, discussing issues such as immigration and the candidates in the district. Allen's primary goal is to listen, not to persuade, as she records voice-to-text summaries of her interactions. Ground Truth, launched by the progressive grassroots group Swing Left, is using artificial intelligence to analyze thousands of open-ended conversations.
The AI-powered app records Allen's conversations and stores them in a central database for further analysis. This technology helps identify unique concerns and issues that may not have been on traditional canvassing questionnaires, such as worries about a rusting water tower in a community. Ground Truth has dispatched over 3,000 volunteers to over 50 House districts across 23 states since January.
Despite its ambitious goal of talking to 500,000 potential voters, the project has only managed to engage over 40,000 as of September. However, Swing Left sees it as a "proof of concept" for the 2028 election and beyond, demonstrating how artificial intelligence can help understand voters and potentially improve Democrats' image as a party that listens to everyone.
Written by urgent.news from Fast Company's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
