For exam-oriented learners, trust in AI tutors hinges on curriculum fit: students abandon mathematically correct solutions that use out-of-syllabus methods¶

Claim · All claims
Evidence · 2 studies · 2 design r2–r3 · q2
Currency · no dated study recorded · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Students immediately stopped reading solutions that used high-school knowledge points for junior-high problems. → trust-quote (2) q2 i? In usability testing, trust collapsed when the AI introduced cotangent in a cube-projection task or vector methods in a coordinate-geometry task. → trust-quote (2)

Evidence¶

trust-quote¶

Harry Feng, Yuan Tian, and Erica Zhao. 2026. From Answer Generators to Reasoning Facilitators: Designing AI Tutors for Mathematical Reasoning in High-Stakes Environments. https://arxiv.org/abs/2607.01692 doi:10.48550/arxiv.2607.01692

q2 · i? · design · r2

Interview quote from participant P09 in the field study, illustrating that distrust targeted the method's legitimacy within the exam ecology rather than only the answer's correctness.

"It used a high-school knowledge point. After I saw that I had not learned it, I stopped reading."

trust-quote (2)¶

Harry Feng, Yuan Tian, and Erica Zhao. 2026. From Answer Generators to Reasoning Facilitators: Designing AI Tutors for Mathematical Reasoning in High-Stakes Environments. https://arxiv.org/abs/2607.01692

q2 · i? · design · r3

Usability study finding (5 participants, think-aloud protocol) mapped in Table 1 to the Exam-Format Layer design change; out-of-syllabus methods can also fail to earn process credit.

"Trust collapsed when the AI introduced cotangent (cot) in a cube-projection task or vector methods in a coordinate-geometry task."

Discussion¶

Related Claims¶

• Review reports a field experiment in which GPT access improved supported practice but was followed by poorer unaided test performance, mitigated by a guarded tutor — related
• In TBLT, teachers perceive scaffolding collapse, a shortcut paradox, invisible inequity, and task realism corrosion — related
• AI-mediating is a newly observed orchestration type in which the teacher concretizes or supplements correct AI feedback for students in curriculum context (12 of 46 episodes) — related
• Mathematics teachers critique ChatGPT-generated generative learning lessons for poor fit to students, boredom, preparation time, missing student-facing materials, mathematical flaws, and required materials — related