Beliefs about accuracy shape confidence attributions to humans and artificial agents

Abstract To effectively communicate and collaborate with others, we must monitor not only other people’s cognitive states (e.g., what someone thinks or believes) but also their metacognitive states (e.g., how confident they are in their beliefs). While humans routinely share confidence, either explicitly (e.g., “I am sure”) or implicitly (e.g., via response times), metacognitive capabilities are still developing in artificial intelligence (AI), raising the question of how humans attribute confidence to AI systems. In seven pre-registered experiments (post-exclusion N s = 113, 109, 56, 59, 52, 60, 57), participants observed human and AI agents make perceptual choices and reported how confident the observed agent seemed in each choice. Overall, attributions of confidence were sensitive to observed behaviour (e.g., task difficulty, accuracy, and response times), but also agent type: observers consistently overestimated the confidence of AI agents compared to humans—even when their behaviour was identical. This illusion of greater confidence in AI decisions was robust across behavioural profiles, agent descriptions, and decision-making domains (visual perception, general knowledge) but was reduced in more subjective decisions (emotion categorisation). An experimental manipulation further showed that illusions of confidence are rooted in prior beliefs about the agents’ capabilities. Together, these investigations of metacognitive attributions reveal a powerful illusion of confidence in artificial systems and highlight a central role for attributions of metacognitive states in human-human and human-AI interactions.