AI Could Strengthen or Undermine Democracy, Blueprint Warns

A new blueprint from the Office of Eric Schmidt argues that AI is becoming the primary interface for democratic participation, with design choices already underway determining whether it strengthens or further strains institutions. The analysis identifies three transformations: how people know (AI-mediated search and synthesis shaping beliefs), how they act (personal AI agents conducting research, drafting communications, and lobbying on users' behalf), and how they engage collectively (AI agents and humans participating in the same forums, potentially producing unanticipated outcomes). The authors, Andrew Sorota and Josh Hendler, warn that without deliberate design, AI could replicate social media's polarization and radicalization risks, but also highlight promising research: AI-generated fact checks on X achieved cross-partisan credibility, AI mediators helped citizens find common ground, and AI models can reduce polarization. They call for new democratic infrastructure—technological and institutional—including truthful model outputs, faithful user representation by agents, identity verification for humans and proxies, and policies to harness AI for responsive governance.