TRUST IN AI SYSTEMS: A SOCIAL-PSYCHOLOGICAL INVESTIGATION OF HUMAN–AI COLLABORATION

Abstract

Trust is a pivotal element in effective human–AI collaboration, influencing whether people adopt, use, or reject AI systems. This research work presents a comprehensive investigation of trust in AI systems from social-psychological and technological perspectives. We examine theoretical foundations of trust, comparing human-to-human trust paradigms with trust in artificial intelligence (AI) agents. Key factors that shape trust in AI are analyzed, including attributes of AI systems (e.g. performance, transparency), human user characteristics (e.g. personality, prior experience), and contextual dynamics (e.g. task risk, team environment). We further explore how trust develops and evolves during human–AI interactions, highlighting phenomena such as overtrust, undertrust, and trust calibration over time. Practical implications are discussed, focusing on strategies to build and maintain appropriate trust through explainable and reliable AI design, user education, and organizational policies. The outcomes of trust – such as user acceptance, reliance on AI recommendations, satisfaction, and collaboration effectiveness – are synthesized from recent research findings. Finally, we outline current challenges (like measuring trust and addressing cultural differences) and future research directions to foster trustworthy AI and optimize human–AI teamwork. Our findings underscore that calibrated trust is essential to harness the full potential of AI while safeguarding human agency and collaboration efficacy.