New Study Reveals How US Users Choose AI Assistants: ChatGPT Dominates, Claude Wins Trust

A new academic study from Jennifer Zou (arXiv, July 2026) provides the first large-scale, weighted analysis of how US adults choose and use AI assistants. Based on a representative sample of 1,999 users, the research reveals a market that is concentrated but highly differentiated. ChatGPT dominates as the primary assistant for 58% of users, followed by Gemini at 25%, yet smaller platforms hold defensible task niches. Claude, despite only a 7% overall share, captures a third of all coding tasks—a clear sign that task allocation is organized by platform strengths rather than user demographics. Technical use also falls steeply with age, suggesting a generational divide in advanced AI adoption.

Trust and privacy emerge as critical differentiators. The study finds that trust is earned through direct use, not reputation: Claude users rate it as the most trustworthy in every head-to-head comparison, and the gap between its users' and non-users' ratings is the largest of any platform. On privacy, concern is near-universal but action is gated by knowledge, not worry. In a choice experiment, users valued preventing human access to their conversations at $11.20/month—far more than blocking model access—and this valuation rises with task sensitivity. These findings have immediate implications for platform strategy, trust-building, and privacy feature design.