Trusting Your AI Agent Emotionally and Cognitively: Development and Validation of a Semantic Differential Scale for AI Trust

Trust in AI is both emotional and cognitive: The research demonstrates that people's trust in AI agents isn't purely rational—emotional responses (affective trust) and logical assessments (cognitive trust) work together to shape overall trust levels, challenging the field's historical focus on cognitive factors alone.

A new measurement tool has been validated: The researchers developed and tested a 27-item semantic differential scale specifically designed to measure both emotional and cognitive dimensions of AI trust, providing a validated tool that can be used across different AI agent contexts.

Emotional and cognitive trust interact dynamically: The study found that affective and cognitive aspects of trust don't operate independently—they interact with each other and collectively influence how much people trust AI agents, suggesting that building trustworthy AI requires attention to both dimensions simultaneously.

LLM-powered conversational agents warrant special attention: The research highlights that newer, human-like language model-based AI agents create unique trust dynamics that differ from previous AI systems, making it important to study how these increasingly sophisticated agents build trust through emotional and cognitive pathways.

Implications for responsible AI design: The findings suggest that developers and organizations deploying AI agents should consider both emotional and cognitive factors when designing for trustworthiness, rather than focusing solely on technical reliability or transparency metrics.