Downloads: 2
Research Paper | Computer Science and Information Technology | Volume 15 Issue 8, August 2026 | Pages: 1642 - 1644 | India
Trust on Trial: How AI Errors Reshape Human Confidence in Technology
Abstract: Artificial intelligence (AI) systems increasingly recommend decisions in domains such as healthcare, hiring, and finance, yet these systems are not infallible. How users calibrate their trust and reliance on such systems after witnessing errors-and whether that calibration holds up over repeated exposure-remains only partly understood. This paper proposes a longitudinal, mixed-methods study examining how trust and behavioural reliance evolve across repeated interactions with an AI recommendation system that makes occasional, controlled errors. We review prior work on trust in automation, algorithm aversion, and trust repair, and we outline a research design that varies error timing, error severity, and the presence of explanations across multiple interaction sessions. We propose measuring both self-reported trust and behavioural reliance (e.g., override frequency, response latency) to capture the gap that often exists between what users say and what they do. We discuss expected patterns- including early over-trust, a sharp drop after a first visible error, and partial but incomplete recovery- along with implications for the design of AI-assisted decision support systems.
Keywords: human-AI interaction, trust calibration, algorithm aversion, AI-assisted decision making, reliance, error recovery
How to Cite?: Shivam Kumar, Shinty P K, "Trust on Trial: How AI Errors Reshape Human Confidence in Technology", Volume 15 Issue 8, August 2026, International Journal of Science and Research (IJSR), Pages: 1642-1644, https://www.ijsr.net/getabstract.php?paperid=SR26820131826, DOI: https://dx.doi.org/10.21275/SR26820131826