The illusion of AI competence: A conceptual review of cross-dimensional miscalibration in generative AI use

Information from the abstract

This article develops a conceptual account of the illusion of AI competence: a calibration failure in which gains in operating generative AI systems are misread as gains in the capacity to evaluate their outputs and in underlying domain knowledge. Drawing on cognitive offloading, the illusion of explanatory depth, the extended mind hypothesis, and research on automation bias and machine heuristics, the article specifies three analytically distinct dimensions of AI competence: operational (AI-O), evaluative (AI-E), and domain-task (DT), arguing that the illusion arises from cross-dimensional inference under asymmetric feedback. A structured but non-exhaustive conceptual review, organized into eight search arms including a deliberate disconfirming arm, maps the current evidence base and identifies where it is thin or contested. The resulting framework is presented as a directed causal model with three moderators, accompanied by two falsifiable implications concerning interface transparency and feedback disruption. The article concludes that the phenomenon is best understood as design-contingent rather than inevitable, with implications for AI literacy curricula and interface design.

Why this record is monitored

This record has an Impact Signal of 79/100 based on recency, source, collaboration, and bibliographic signals. It prioritizes monitoring and is not a judgment of research quality.

Related topics: Human-Automation Interaction and Safety · Ethics and Social Impacts of AI · Cybernetics and Technology in Society