Beyond Risk Reduction: Vigilant Trust in Artificial Intelligence Based on Evidence from China

Area of Science:

• Social Sciences
• Technology and Society
• Artificial Intelligence Ethics

Background:

• Public trust in artificial intelligence (AI) is commonly believed to reduce perceived risks, thereby increasing acceptance.
• This assumption overlooks the complex relationship between trust, risk perception, and AI adoption.

Purpose of the Study:

• To challenge the assumption that trust in AI solely reduces perceived risks.
• To introduce and explore the concept of 'vigilant trust' in artificial intelligence.
• To examine how different dimensions of trust influence perceptions of benefits and risks, and subsequently, AI acceptance.

Main Methods:

• A nationally representative survey of 10,294 Chinese adults was conducted.
• The study analyzed four dimensions of trust: trusting stance, competence, benevolence, and integrity.
• Statistical analysis examined the relationships between trust dimensions, perceived benefits, perceived risks, and AI acceptance.

Main Results:

• All dimensions of trust consistently enhanced perceived benefits, the primary driver of AI acceptance.
• Trust exhibited differentiated effects on risk perception: benevolence decreased it, while a trusting stance increased both benefit and risk perceptions.
• Perceived risks did not consistently deter AI acceptance; in some cases, they were positively associated with adoption willingness.

Conclusions:

• Trust in AI does not necessarily diminish risk awareness; 'vigilant trust' coexists with and can intensify attention to potential harms.
• AI acceptance is complex, influenced by perceived benefits and a nuanced understanding of risks, not solely risk reduction.
• Findings suggest a need for governance and communication strategies that acknowledge vigilant trust and sustained risk awareness in AI adoption.