Synchronous AI suggestions shape subsequent writing, offering new ideas but risking influence over the writer's voice and argument¶

Claim · All claims
Evidence · 1 study · causal r2 · q2
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 1 of 1 entry) · not reviewed by a person

Subclaims¶

q2 i? Participants reported that synchronous comments could change their train of thought and guide subsequent writing, sometimes against their wishes. → 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 · r2

In the thematic analysis of interviews (§5.1.3), participants valued suggestions when stuck but worried about lost voice; P14 said "if I incorporate the first comments, then I feel like my train of thought would change according to its response".

"if I incorporate the first comments, then I feel like my train of thought would change according to its response"

Discussion¶

Learner Variables¶

• Writing Skills — outcome: instruction changes it

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
• Synchronous collaboration with an artificial agent amplifies contextual misalignment because the agent only sees ongoing output, not the writer's planning process — related
• Synchronous autonomous AI suggestions increased perceived writing efficiency but also caused distraction and mental overload during writing — related
• AI support shifts perceived attribution of ideas and text toward the AI, most strongly for AI-generated drafts, and an AI draft based on the writer's own outline contributed more ideas than expected (27%). — related
• Three multilingual international students critically evaluated AI feedback, recognized peer-AI agreement, and maintained their voices while revising — related
• Planning-stage AI ideas were often redundant with or rejected in favor of writers' own ideas: idea attribution was 17.8% among users versus 2.9% among non-users, but ownership did not differ between users and non-users within ai plan. — related
• Students reported text supports structured, iterative planning while some valued voice for getting initial ideas out — related
• An agentic cursor increased participants' understanding and predictability of the AI agent's actions during collaborative writing — related
• Adding a second writing agent increased reported cognitive load, and writing confidence did not increase monotonically with support, with the Single condition reporting the lowest confidence — related