Participants rated the arithmetic-error peer agent more harshly than the conceptual-error peer agent, calibrating trust by error observability rather than severity¶

Claim · All claims
Evidence · 1 study · causal r? · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Across nearly every rated dimension, the arithmetic-error peer Alice was rated more harshly than the conceptual-error peer Charlie, despite both being designed as error-prone peers. → Harsh Kumar 2026

Evidence¶

Harsh Kumar 2026¶

Harsh Kumar, Zi Kang (Jace) Mu, Jonathan Vincentius, Ashton Anderson. (2026). Human Thinking under Plural LLM Assistance: Mathematical Problem Solving and Open-Ended Writing. Working Paper, July 2026, Toronto, Canada. https://arxiv.org/abs/2604.02677 doi:10.48550/arxiv.2604.02677

q2 · i? · causal · r?

Post-survey agent ratings (competence, warmth, helpfulness, trustworthiness, 1–5 Likert) in Experiment-1; the authors interpret the asymmetry as trust calibrated to the observability of errors rather than their severity, an interpretation rather than a tested result.

"participants rated Alice (strong concepts, arithmetic errors) more harshly than Charlie (accurate arithmetic, conceptual errors) on nearly every dimension, despite both being designed as error-prone peers."

Discussion¶

Related Claims¶

• Teachers of color are judged more harshly in evaluations: within the same schools, Black teachers were 50 percent more likely than white teachers to receive a low rating. — related
• Degree of anthropomorphism attributed to the agent was shaped by both interaction modality and, within audio conditions, voice accent — related
• Role ambiguity for a Black-accented agent destabilized trust, culminating in breakdown after the agent inaccurately tracked group decisions — related
• Speakers rate their own speeches more harshly than independent judges do — related
• Expert evaluators rate PIVOT higher than TheoremExplainAgent across nearly all instructional dimensions, with the largest gains in Physics and Computer Science — related