Abstract
Metacognition, or "thinking about thinking," is a pivotal ability in human decision-making. As artificial intelligence (AI) evolves, a question emerges: “What occurs when this capability is integrated into AI systems?” This ongoing research investigates the implications of embedding metacognition in AI advisors on user advice utilization. Study 1 assesses users' perceptions of metacognition in AI advisors compared to human experts. Study 2 analyzes how different levels of AI metacognitive abilities (high vs. low) impact AI advice utilization, specifically through perceived AI expertise and system causability. Study 3 gauges how decision type—ill-structured vs. well-structured— and perceived user’s expertise affect interactions with AI metacognition and advice incorporation. Through online experiments, participants interact with advisors, offer initial judgments, receive advice, and make final decisions. Grounded in metacognition theory and the Judge Advisor System (JAS) framework, this research-in-progress aims to elucidate the intricate dynamics between AI metacognition and users' advice utilization.
Recommended Citation
Mehri, Bita and Ghasemaghaei, Maryam, "Does AI Metacognitive Ability Lead to Higher AI Advice Utilization?" (2023). Proceedings of the 2023 Pre-ICIS SIGDSA Symposium. 9.
https://aisel.aisnet.org/sigdsa2023/9