Participants remain socially sensitive to conversational stance even while categorical identity inference fails¶

Claim · All claims
Evidence · 1 study · causal r2 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? Supportive agents were rated substantially more supportive and markedly less conflictual than contrarian agents, and trust varied strongly by stance within AI targets. → Lixiang Yan 2026

Evidence¶

Lixiang Yan 2026¶

Lixiang Yan, Yueqiao Jin, Xibin Han, Dragan Gasevic. (2026). Socially fluent AI decouples conversational signals from source identity in online interaction. https://arxiv.org/abs/2605.23426 doi:10.48550/arxiv.2605.23426

q3 · i? · causal · r2

Manipulation checks among 685 AI targets confirmed the stance manipulation: supportive agents rated more supportive (M = 5.77 vs 4.47) and less conflictual (M = 2.51 vs 4.25) than contrarian agents. Within AI targets, trust varied by stance (supportive M = 5.57, n = 344; contrarian M = 4.87, n = 341).

"In manipulation checks among AI targets (𝑛 = 685), supportive agents were rated as substantially more supportive than contrarian agents (supportive: 𝑀 = 5.77, 𝑆𝐷 = 1.24; contrarian: 𝑀 = 4.47, 𝑆𝐷 = 1.91) and as markedly less conflictual (supportive: 𝑀 = 2.51, 𝑆𝐷 = 1.73; contrarian: 𝑀 = 4.25, 𝑆𝐷 = 1.85)"

Discussion¶

Related Claims¶

• Participants fail to distinguish undisclosed AI teammates from human teammates above chance in synchronous text-based group interaction — related
• Representational similarity analysis shows identity judgments are organised around subjective impressions, orthogonal to diagnostic cue structure — related
• Minimal human-like cues suffice to trigger automatic social reactions to embodied agents, though the role of visual social cues remains contested — related
• General-purpose LLMs show sycophancy: AI responses affirmed users substantially more than human responses, increasing trust and reducing willingness to take responsibility — related