Agent perception followed a U-shape across variants: the mixed temporally-humanlike/visually-nonhumanlike Variant 2 was rated least positively on partnership and agent-perception measures¶

Claim · All claims
Evidence · 1 study · causal r1 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? PMQ and ASA scores were lowest for Variant 2, with Variant 1 and Variant 3 rated more positively and minimally different from each other. → Michael Yin 2026

Evidence¶

Michael Yin 2026¶

Michael Yin, Angela Chiang, Samuel Rhys Cox, and Robert Xiao. (2026). "It Felt a Bit Eerie": Exploring Humanlike Interactions During Collaborative Writing with an Artificial Agent. arXiv. https://arxiv.org/abs/2605.24729 doi:10.48550/arxiv.2605.24729

q2 · i? · causal · r1

Exploratory analysis of PMQ and ASA questionnaire data (n=48, 16 per variant) with Mann-Whitney tests found "Variant 2 generally scored lower than the other variants on all factors"; the U-shape was strongly suggested by descriptive statistics and somewhat supported by tests.

"Our analysis shows that Variant 2 generally scored lower than the other variants on all factors of communicative competence and dependability, humanlikeness in communication, and communicative flexibility, whereas Variant 1 and Variant 3 showed minimal significant differences."

Discussion¶

Related Claims¶

• Humanlike AI co-writing (real-time suggestions and an agentic cursor) evoked feelings of surveillance, judgment and eeriness, and the moving cursor distracted writers — related
• Participants generally felt humanlikeness should not be a design goal for a writing-support agent used as a tool — related
• An agentic cursor increased participants' understanding and predictability of the AI agent's actions during collaborative writing — related
• U-shaped dynamics: partnership-vigilance and offloading-transformation relationships are convex, with benefits accelerating only beyond a threshold of engagement — related