AI Assistance and Metacognitive Calibration
Document Type
Dissertation
Degree Name
Doctor of Philosophy (Ph.D)
Department
Psychology and Special Education
Date of Award
Spring 5-1-2026
Abstract
The present study examined how access to artificial intelligence (AI) assistance influences students’ metacognitive judgments and introductory statistics learning outcomes. Research shows that external assistance (e.g., hints) could lead individuals to overestimate their own abilities by misattributing externally supported success to internal competence (Fisher & Oppenheimer, 2021). After viewing a statistics lecture, participants were randomly assigned to one of three conditions: Always AI access, Delayed AI access, or No AI. Participants completed practice problems in the assigned condition, made estimations of future performance (EFP) on a test without external assistance, took a learning outcome assessment, and completed measures of individual differences and demographics. Relative to the No AI condition, the Always AI condition reported significantly lower EFP, whereas the Delayed AI condition did not. Although practice with AI assistance had positive, but small, effects on both EFP and actual learning, practice with AI failed to improve actual learning. Instead, AI assisted learning harmed the learning outcome performance. Variations in one’s prior knowledge of the topic, math anxiety, and intellectual value were significant covariates of the present results. The current study did not observe the overconfidence reported in the previous research. The difficulty space created by delayed AI access failed to show significant benefits in learning.
Advisor
Shulan Lu
Subject Categories
Education
Recommended Citation
Ballotti, Reynolds John, "AI Assistance and Metacognitive Calibration" (2026). Electronic Theses & Dissertations. 1347.
https://lair.etamu.edu/etd/1347
