Beyond Individual Agency: Designing LLM-Assisted Tools that Foster Collaborative Journalistic Deliberation

Building on findings from "Designing for Agency in LLM-Infused Writing Support Tools for Science Journalism," this research shifts focus from individual journalist agency to collective agency within newsroom teams. It proposes designing LLM-assisted writing tools that scaffold and enrich collaborative deliberation and distributed decision-making. Key features include transparent provenance and audit trails showing how story decisions evolved collaboratively; interactive deliberation prompts encouraging dissent, counterarguments, and ethical discussions; role-specific agency controls allowing different newsroom roles to adjust LLM interventions; and embedded consensus-building workflows that summarize agreements, disagreements, and unresolved debates. This approach challenges the norm of maximizing automation for individual productivity by prioritizing collaborative processes and meta-cognition. It reframes AI as a facilitator of healthy journalistic debate and distributed editorial agency, potentially improving transparency, reducing groupthink, and modeling accountability in AI-human editorial workflows, especially in science journalism.

References:

1. Designing for Agency in LLM-Infused Writing Support Tools for Science Journalism. Sachita Nishal, Mina Lee, Nicholas Diakopoulos, Jennifer Wortman (None).