Revisiting Trust in the Era of Generative AI: Factorial Structure and Latent Profiles

Trust in AI involves both rational and emotional dimensions: The research identifies four key components of trust in generative AI systems—Affective Trust, Competence Trust, Benevolence & Integrity, and Perceived Risk—showing that people's confidence in AI goes beyond just whether it works reliably to include how it makes them feel and whether they believe it has their best interests in mind.

GenAI systems require a different trust framework than traditional tools: Because generative AI systems converse and collaborate with users rather than simply processing information, the paper argues that measuring trust in these systems requires moving beyond traditional functionality-focused metrics to capture the social and emotional dimensions that matter when AI feels more like a partner than a tool.

Different user groups have fundamentally different trust profiles: The study identified six distinct trust profiles among users, revealing that people don't fall into a simple "trusting" or "distrusting" category. Instead, individuals combine affective and competence-based trust in different ways, suggesting that one-size-fits-all approaches to building AI trust may be ineffective.

Cultural differences shape how people trust AI: The research drew on large-scale surveys from both China and the United States, indicating that trust in generative AI is culturally sensitive and that responsible AI deployment requires understanding these cultural variations rather than applying universal trust-building strategies.

A validated measurement tool is now available for researchers and practitioners: The Human-AI Trust Scale (HAITS) provides organizations with a scientifically validated method to assess how different user groups experience trust in generative AI systems, enabling more targeted approaches to responsible AI development and deployment.