Directional Alignment and Narrative Agency in Human–LLM Co-Writing

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Expert, quick

Summary

An investigation into narrative agency in human–LLM creative co-writing reveals an asymmetric influence where humans drive narrative innovation and direction, while LLMs primarily elaborate on human-introduced elements and sustain coherence. Using a new corpus of human–LLM co-written stories, researchers applied sentiment and semantic modeling to quantify affective alignment and semantic novelty in turn-taking. Results show human turns introduce greater semantic novelty and shape subsequent developments. LLM contributions predominantly elaborate on human-introduced elements. Sentiment alignment is asymmetric but bidirectional, with LLMs exhibiting stronger turn-level emotional adaptation than humans, though both agents track each other's emotional valence. LLMs also show an independent tendency towards more positive emotional baselines.

Key takeaway

For NLP Engineers designing co-creative AI systems, these findings suggest focusing human input on core narrative innovation. You should configure LLMs to act as adaptive amplifiers, elaborating on human-introduced elements and maintaining emotional coherence. This division optimizes co-writing by utilizing human creativity for direction and LLMs for sustained narrative development, enhancing collaborative story generation.

Key insights

Humans drive narrative innovation in co-writing, while LLMs adaptively elaborate and maintain coherence.

Principles

• Human turns introduce greater semantic novelty.
• LLMs primarily elaborate on human-introduced elements.
• LLMs show stronger emotional adaptation in co-writing.

Method

Quantify affective alignment and semantic novelty using sentiment and semantic modeling on a human-LLM co-written story corpus, applying directional measures to assess narrative progression.

In practice

• Design LLM co-writing for human-led innovation.
• Utilize LLMs for narrative elaboration and coherence.
• Employ LLM emotional adaptation for story flow.

Topics

• Narrative Agency
• Human-LLM Co-Writing
• Sentiment Modeling
• Semantic Novelty
• Creative AI
• Large Language Models

Best for: Research Scientist, AI Scientist, NLP Engineer

Related on AIssential

• Improving Collaborative Storytelling with a Multi-Agent Framework Based on Large Language Models — An iterative LLM Writer-Editor framework significantly improves collaborative storytelling quality, even with few refinement steps.
• Automated Creativity Evaluation of Language Models Across Open-Ended Tasks — The framework offers a scalable, domain-agnostic method to evaluate LLM creativity by separating measurement from task.
• StoryCoder: Narrative Reformulation for Structured Reasoning in LLM Code Generation — Narrative reformulation of code problems significantly improves LLM code generation by structuring context.
• Exploring conversational AI and poetry but not as we know it — LLMs challenge traditional notions of language, creativity, and authenticity, particularly in fields like poetry.
• To LLM, or Not to LLM: How Designers and Developers Navigate LLMs as Tools or Teammates — LLM adoption is a sociotechnical positioning problem, not merely a technical evaluation of capability.
• How do ML researchers actually use AI tools to improve their writing? [D] — LLMs offer efficiency in technical writing and coding but risk diminishing cognitive skills and originality if over-relied upon.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.