Congratulations to Halfdan Nordahl Fundal, Rebekah Baglini, & collaborators Yuri Bizzoni, Charlotte Bilde and Ida Bække Johannesen on their paper "Humans Introduce, Models Elaborate: Asymmetric Narrative Agency in Human-LLM Co-Writing”
Accepted as a Main Conference Paper at EMNLP 2026 in Budapest. The paper has also been nominated for a best paper award.
EMNLP is one of the leading conferences in Natural Language Processing. This year, the conference received a record 17,669 submissions, with only 15.4% of Main Conference Papers being accepted.
Their paper, produced as part of Rebekah Baglini’s work package at the TEXT Center, explores how humans and large language models collaborate when writing stories together. By comparing Human-Human, Human-AI, and AI-AI storytelling, the researchers found that humans are more likely to introduce new ideas that shape the story, while AI models tend to build on and develop those ideas. Rather than acting as equal co-authors, LLMs appear to function as narrative amplifiers that help expand and stabilize human contributions - but are not the primary drivers of creativity.
You can read a preprint of the accepted paper here: