A warning message about potential AI mistakes increases seventh-graders' hint requests in a math intelligent tutoring system¶

Claim · All claims
Evidence · 1 study · causal r2 · q3
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 1 of 1 entry) · not reviewed by a person

Subclaims¶

q3 i? Students shown a warning that the pedagogical agent can make mistakes requested hints more frequently than students shown a neutral message, even though system behavior was identical. → Tomohiro Nagashima 2026

Evidence¶

Tomohiro Nagashima 2026¶

Tomohiro Nagashima, Mirella Hladký, and Vera Rief. (2026). Warning About AI Fallibility Increases Help-Seeking in an Intelligent Tutoring System. https://arxiv.org/abs/2606.03822 doi:10.48550/arxiv.2606.03822

q3 · i? · causal · r2

Classroom experiment at a secondary school in Tokyo with 252 seventh-grade students randomly assigned to Mistake or NoMistake tutor versions; linear regression on log data showed "students in the Mistake condition requested hints more frequently than those in the NoMistake condition" (β=−0.33, t(235)=−2.33, p=.02).

"Results showed that students in the Mistake condition requested hints more frequently than those in the NoMistake condition (β=−0.33,t(235) =−2.33,p=.02)."

Discussion¶

Related Claims¶

• Hint takers spend less time on a question than learners who attempt it, regardless of attempt outcome — related
• Not requesting any hints in at least one problem is related to greater wheel-spinning, consistent with help avoidance — related
• The AI-fallibility warning did not significantly change error rate or time spent per problem-solving step — related
• Adaptive students request significantly fewer on-demand hints and spend longer on a step before requesting help than Control students — related
• In Study 2, both general and specific warnings at least marginally reduce attitude change, with no significant difference between them — related
• Undergraduate computing students report help-seeking avoidance and anxiety, with 62.5% sometimes avoiding asking for help even when needed and 75% reporting anxiety when not understanding a topic — related
• Most students (87.6%) prefer attempting to solve problems independently before seeking support, managing uncertainty privately before engaging formal support — related
• Warning labels alone did not improve performance or reduce reliance on GenAI advice, and students sometimes adopted incorrect AI answers — related
• Error rates per completed problem-solving step did not significantly differ between conditions — related
• Students receiving personalized problems gamed the tutoring system less during the personalized unit, with no difference in a later unit — related
• For some students, AI-generated feedback reduces the perceived personal risk of exposing weaknesses, functioning as a psychologically safe space for feedback seeking — related
• The Help Tutor improved students' in-tutor help-seeking behavior, but only for a subset of help-seeking action types — related
• Mindful-condition students completed fewer problem-solving steps and requested fewer hints per step than Cognitive-condition students — related