HUMAN METACOGNITION IN AI-ASSISTED DECISION- MAKING

Artificial Intelligence (AI) fundamentally transforms decision-making. Its increasing accessibility and growing capabilities make it more and more feasible for humans to seek guidance in both professional and personal contexts. However, the responsibility to accept or reject AI recommendations ultimately rests with the human user, a process that relies heavily on metacognitive skills. Despite the growing integration of AI into decision-making processes, evidence on its effects on human cognition, especially metacognitive processes, remains limited. We assume that AI assistance reduces users’ metacognitive monitoring during decision-making, fostering overconfidence in AI capabilities and impairing metacognitive oversight in subsequent decisions that do not depend on AI. Through an experimental approach, we investigate how human-AI collaboration affects metacognition, with the goal of informing responsible use and design in AI systems to uphold human agency in decision-making.