The platform's out-of-scope audit protocol is designed to neutralize student distrust of AI error by rewarding verified hallucination reports¶

Claim · All claims
Evidence · 1 study · theoretical r3 · q1
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q1 i? The author argues that granting students authority to challenge AI premises, with instructor adjudication on a four-point scale, structurally addresses student anxieties about algorithmic error and builds trust. → Ilya Mikhelson 2026

Evidence¶

Ilya Mikhelson 2026¶

Ilya Mikhelson. (2026). The Theoretical Foundation of Socratic Tests: Dynamic, Multimodal, Conversational Examinations. https://arxiv.org/abs/2607.29624 doi:10.48550/arxiv.2607.29624

q1 · i? · theoretical · r3

Design argument (Section 5.4), not a tested result: the author states the mechanism "builds deep trust in the assessment modality by guaranteeing that human oversight (the instructor's audit) remains the ultimate arbiter of algorithmic fairness," citing literature that perceived risk deters adoption. No trust measures are reported.

"By granting students the explicit authority to challenge the AI’s premises and forcing a pivot in questioning when a challenge is issued, the platform structurally empowers the learner."

Discussion¶

Related Claims¶

• Participants articulated four interrelated arguments for preserving human instructor authority in assessment: knowledge of students, dialogue, institutional legitimacy, and safeguarding against AI error — related
• Of 100 human-audited claims, 39 are fully verifiable, 55 partially verifiable, and 6 not verifiable, with main responses showing stronger provenance than follow-ups — related
• Of 30 verified hallucinated references, 13 appeared entirely fabricated and 17 were hybrid references combining a real title with fabricated authorship or metadata — related
• Verified hallucinated references at SIGCSE TS increased from 3 in the 2025 proceedings to 17 in the 2026 proceedings, appearing in 2.3% of 2026 papers — related
• AI hallucination can be turned into a pedagogical resource by making verification and collaborative fact evaluation integral to AI-supported learning — a broader claim this one bears on
• In a ten-request live textbook audit, citation integrity coexisted with substantive error: every citation resolved to its source, yet answers included retrieval misses and contradictions — related
• Three recurring AI failure modes arose in this project-based learning context: plausible-but-incorrect code, missing specialized knowledge, and limited long-term context — related
• Conservative manual verification identified 30 hallucinated references across 14 papers in 2025 and 2026 computing education venues — related
• The member check did not resolve the dissensus: the distrust item (C4) was the weakest-endorsed confirmatory item, with participants reframing what distrust meant — related
• In a small case study, students given ChatGPT feedback valued its immediacy and alternative analytical perspectives, but one distrusted its answers as 'often made up' for complex data analysis — related