Intrinsic Honesty and Trust in AI

This dataset is used in the paper "From Intrinsic Honesty to Trust in Artificial Intelligence (AI): How Moral Integrity
Influences Trust in AI Systems".

This paper uses a repeated trust game to explore trust dynamics in artificial intelligence (AI) systems. The experiment assessed how trust is formed and evolves with AI systems exhibiting different personalities. Participants engaged in repeated tasks, allowing the distinction between initial trust and trust over time. Findings indicate that socially oriented AI, with consistent and rewarding behavior, elicited the highest trust levels but was least reciprocated with trust compared to competitive and exploratory AI systems. Trust was also sensitive to perceived fairness, since participants were less trusting when AI had the same initial endowment. Trust recovery after a breach proved challenging regardless of the initial level of trust. The findings offer practical insights for designing trustworthy AI systems and informing AI governance policies. The study advances our understanding of trust in AI, offering actionable guidance for AI deployment in trust-sensitive environments.