Prompting for the Public Mission: How Journalists Shape AI Tools

This work-in-progress explores how prompts operate as a mediating layer through which
journalists are enabled to shape AI tools to their public service mission. Newsrooms deploying AI
editing tools provide a critical site where traditional journalistic practice and the machine logics of
third-party AI services meet. Journalists do not merely delegate tasks to AI tools; rather, they
engage in a communicative loop, negotiating model outputs through strategic prompting. In
newsroom practices, prompt templates enable journalists to stabilize this interaction and not only
act as users but also as co-creators of the tool’s operational logic and affordances. In this way,
prompting emerges as a “form of programmability” (Rieder, 2024) and, consequently a “form of
power” (Amoore et al, 2025), granting participating journalists both the freedom and responsibility
to translate editorial intentions and public service principles into machine-readable instructions.
Drawing on an ongoing case study of the deployment of an AI editing tool at a European public
service media (PSM) organisation, I examine how communicative AI reshapes journalistic practices
by introducing prompting as a new form of work and communication. In this crucial phase of
reorganisation, the question arises: To what extent is the public service mission of PSM
journalists reflected in the learning and doing of prompting? Based on ethnographic
observations, interviews with developers and prompt writers, and thematic analysis of prompt
archives, I investigate how prompting is understood, taught, embedded, and distributed in current
newsroom routines. For this purpose, both the themes and structures of prompts themselves and
the social processes of their creation are examined, through training seminars, prompt archives,
and affordances of working tools.