期刊:Advances in computational intelligence and robotics book series [IGI Global] 日期:2026-05-07卷期号:: 93-116
标识
DOI:10.4018/979-8-3373-7320-1.ch004
摘要
As generative artificial intelligence (AI) is adopted in knowledge management, its propensity to hallucinate, giving rise to confident fabrications, threatens not only facts but the foundations of human trust. This chapter argues that trusting AI is not a technical or explainable problem, but an existential one. Using phenomenological insights from Husserl, Heidegger and Merleau-Ponty, it is shown how hallucinations disturb users' sense of intelligibility, accountability and epistemic agency. Empirical studies show that leadership, culture and design count more than system features for building trust. Existing frameworks like MAST and POLARIS are criticized for ignoring the affective dimension, and a new model of trust as “knowing-with” AI is developed, using narrative repair, co-decision architecture and joint stewardship. Truthful AI, it is claimed, needs existential coherence.