
This project investigates how human trust is formed and maintained when collaborating with generative AI agents in decision-making tasks. As generative AI becomes integral to everyday and professional contexts, understanding when and why users trust these systems is crucial for responsible and effective design. Focusing on cognitive and calibrated trust, the project examines how agent qualities (such as explanation strategies, uncertainty handling, and error repair) and environmental factors (like task difficulty and time pressure) influence users’ trust over time.
Through a series of empirical studies, we develop and evaluate a trust-aware generative AI assistant that adapts its explanations and communication style to user context. The goal is to enable human-AI collaborations where users’ trust appropriately aligns with the agent’s actual capabilities, supporting transparency, reliability, and meaningful cooperation between humans and generative systems.




