Imperfectly Cooperative Human-AI Interactions: Comparing the Impacts of Human and AI Attributes in Simulated and User Studies

A research study comparing simulated AI interactions with real human subjects reveals that AI transparency significantly outweighs personality factors in determining interaction quality, with findings diverging notably between pure simulation and actual human experiments across hiring and transactional scenarios.

This research addresses a critical gap in human-AI interaction design by empirically testing how different variables influence outcomes in imperfectly aligned scenarios—situations where humans and AI systems have conflicting objectives. The study's dual methodology of 2,000 simulations paired with 290 human subjects creates a valuable comparison point that exposes the limitations of purely computational models in predicting real-world behavior.

The fundamental insight—that AI transparency dramatically outperforms human personality traits in actual human-AI interactions—carries significant implications for AI development practices. Simulations tend to operate under idealized assumptions about how humans process information and respond to AI behavior, often underweighting the trust and psychological comfort that transparency provides. When humans interact with opaque AI systems, particularly in high-stakes scenarios like hiring or financial transactions, uncertainty breeds skepticism regardless of their personal disposition.

This finding challenges current industry assumptions about customizable AI personalities and adaptive systems. Many companies invest heavily in designing AI agents with specific personality attributes, yet this research suggests transparency mechanisms deliver more measurable value. The divergence between simulated and human results indicates that real humans apply contextual reasoning and emotional interpretation that computational models struggle to replicate.

For AI deployment, the implications are practical: organizations should prioritize explainable decision-making and clear communication of AI limitations over personality engineering. The research particularly emphasizes this in transactional contexts where information asymmetry presents ethical concerns. As AI systems increasingly handle consequential decisions affecting human welfare, this transparency-first approach aligns with both user preferences and responsible AI principles.

• →AI transparency proved significantly more impactful than human personality traits in real human-AI interactions, contradicting simulation-based predictions.
• →Purely simulated AI interactions produce measurably different results from human subject experiments, suggesting computational models miss critical human factors.
• →Hiring negotiations and transactional scenarios showed different outcome patterns, indicating context-specific design requirements for human-AI systems.
• →Real humans prioritize explainability and trust signals over customized AI personality attributes in high-stakes interactions.
• →Imperfectly cooperative scenarios expose divergences between theoretical AI design and practical human expectations.