EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

The Effect of Trust in Generative AI on Critical Thinking: The Mediating Role of AI Over-Reliance

Rina Matsumoto, Kenta Ishii, Shota Miyazaki

SOCIAL AND MANAGEMENT SCIENCES RESEARCH JOURNAL · 29.09.2026 · 4 (2) · 166-177

DOI: 10.66238/smsrj.v4i2.77 ↗

Kurz gefasst

Fachartikel in der Zeitschrift SOCIAL AND MANAGEMENT SCIENCES RESEARCH JOURNAL, online erschienen am 29.09.2026. Verfasst von Rina Matsumoto und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Trust is necessary for productive use of generative artificial intelligence (GenAI), yet excessive trust can shift cognitive work from the learner to the system. This study tests whether AI over-reliance mediates the relationship between trust in GenAI and critical-thinking disposition. We conduct a reproducible secondary mediation analysis using a public dataset of 2,614 Bangladeshi university students and construct-level statistics reported in its companion conference study. Trust in AI was strongly associated with cognitive dependence (r = .707). Using standardized ordinary least-squares mediation reconstructed from the published correlation matrix, the trust→dependence path was a = .707; the dependence→critical-thinking path controlling for trust was b = .150; the direct trust→critical-thinking path was c′ = .242; and the indirect effect was ab = .106 (delta-method SE = .018, z = 5.78, 95% CI [.070, .142]). The total trust–critical-thinking association was c = .348, so 30.5% of the observed association was carried through cognitive dependence. Importantly, the indirect path is positive in this self-report dataset, which does not support a simple “more dependence means less critical thinking” account. We interpret this pattern as a calibration paradox: high-trust students may simultaneously report stronger critical-thinking tendencies and stronger reliance, while behavioral studies can still show performance costs from uncritical acceptance. The paper contributes a dual-pathway framework that distinguishes trust, dependence, and verification quality, and derives design implications for GenAI-supported learning.

Zitieren

APA

Matsumoto, R., Ishii, K., & Miyazaki, S. (2026). The Effect of Trust in Generative AI on Critical Thinking: The Mediating Role of AI Over-Reliance. SOCIAL AND MANAGEMENT SCIENCES RESEARCH JOURNAL, 4(2), 166-177. https://doi.org/10.66238/smsrj.v4i2.77

Vancouver

Matsumoto R, Ishii K, Miyazaki S. The Effect of Trust in Generative AI on Critical Thinking: The Mediating Role of AI Over-Reliance. SOCIAL AND MANAGEMENT SCIENCES RESEARCH JOURNAL. 2026;4(2):166-177. doi:10.66238/smsrj.v4i2.77

BibTeX
@article{Matsumoto2026,
  author = {Matsumoto, Rina and Ishii, Kenta and Miyazaki, Shota},
  title = {The Effect of Trust in Generative AI on Critical Thinking: The Mediating Role of AI Over-Reliance},
  journal = {SOCIAL AND MANAGEMENT SCIENCES RESEARCH JOURNAL},
  year = {2026},
  volume = {4},
  number = {2},
  pages = {166--177},
  doi = {10.66238/smsrj.v4i2.77}
}
Kostenlos starten →