Students frequently accepted misleading AI recommendations (overreliance) while rarely rejecting correct ones (underreliance)¶
Claim · All claims
Evidence · 1 study (2 entries) · associationalr1–r2·q3
Currency · newest study 2026 · checked against its sources 2026-10-17 (automated, 2 of 2 entries) · not reviewed by a person
Subclaims¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q3 i? Mean appropriate reliance was 61.77% (SD = 10.78) and task accuracy 61.21% (SD = 10.93). → Griffin Pitts 2026
q3 i? Overreliance averaged 86.03% of misleading trials while underreliance averaged 2.37% of correct trials. → Griffin Pitts 2026 (2)
Evidence¶
Griffin Pitts 2026¶
Griffin Pitts, Neha Rani, and Weedguet Mildort. (2026). Trust and Reliance on AI in Education: AI Literacy and Need for Cognition as Moderators. Preprint, accepted to AIED 2026. https://arxiv.org/abs/2604.01114 doi:10.48550/arxiv.2604.01114
q3 · i? · associational · r2
Descriptive statistics from the Results section of the 432-participant laboratory experiment with 14 Python output-prediction problems; these are descriptive means with no test statistic or effect size printed.
"on average, students appropriately relied on the AI support in 61.77% of opportunities (SD= 10.78,N= 432), and task accuracy was similar in level (M= 61.21%,SD= 10.93)."
Griffin Pitts 2026 (2)¶
Griffin Pitts, Neha Rani, and Weedguet Mildort. (2026). Trust and Reliance on AI in Education: AI Literacy and Need for Cognition as Moderators. Preprint, accepted to AIED 2026. https://arxiv.org/abs/2604.01114
q3 · i? · associational · r1
Descriptive reliance measures from the same Results section: overreliance M = 86.03% (SD = 25.52) and underreliance M = 2.37% (SD = 8.27). These are descriptive percentages; no effect size is printed.
"students frequently followed misleading recommendations on the subset of misleading trials (overreliance: M= 86.03%,SD= 25.52),whilerejectingcorrectrecommendationswasuncommon (underreliance:M= 2.37%,SD= 8.27)."
Discussion¶
Related Claims¶
- Reflection produced selective rather than blanket rejection of AI: alignment with correct advice stayed high while alignment with incorrect advice dropped — related
- Reviewed literature reports risks of overreliance, AI errors in complex tasks, and students' uncritical acceptance of AI-generated outputs — a broader claim this one bears on
- A brief metacognitive reflection prompt reduces acceptance of incorrect AI advice compared with open ChatGPT support alone — related
- ChatGPT recommendation accuracy was comparable between the open-support and reflection conditions, ruling out unequal exposure to erroneous advice — related
- Trust and reliance can diverge, and self-reported reliance measures cannot classify calibration without output-quality ground truth — related
- Reflection improves correspondence between perceived and behavioral reliance on ChatGPT — related
- Students exhibited high trust in GenAI but limited critical evaluation of its outputs — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
overreliance-frequent-underreliance-rare |
title |
Students frequently accepted misleading AI recommendations (overreliance) while rarely rejecting correct ones (underreliance) |
description |
Students frequently accepted misleading AI recommendations (overreliance) while rarely rejecting correct ones (underreliance) |
status |
review |
evidence_strength |
moderate |
newest_evidence |
2026 |
source_checked |
2026-10-17 |
generated |
by process:wiki-ingest on 2026-10-11 |
trust |
unverified — no one has confirmed this page's content |
sources |
2 — griffin-pitts-2026, griffin-pitts-2026-2 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.