Prompting for the Public Mission: How Journalists domesticate AI Tools

This study applies communicative constructivism to conceptualise prompting practices in
newsrooms as communicative action, through which journalists and communicative AI negotiate
knowledge and editorial norms in PSM newsrooms. Prompts are analysed as objectivations of
journalistic knowledge that stabilise and circulate workflows and task definitions. Thereby
prompting practices are analysed as a form of programmability and power in human–machine
communication (HMC), framing prompts as a mediating layer through which journalistic practices
both configures and is configured by AI infrastructures.

The research is based on an ongoing case study at a European PSM organisation, combining
ethnographic observations from prompt workshops with journalists, the analysis of technical
affordances of the prompt-builder interface and the prompt library, and thematic analysis of
prompts themselves. The study focuses on how prompt templates and associated practices
become objectivated and institutionalised within newsroom routines, attending both to the
structures of prompts and the social processes through which they are created, such as training
seminars, archival practices, and tool affordances.

This research provides an empirically grounded exploration of AI prompting as a communicative
and knowledge-producing practice within PSM newsrooms. By focusing on the mediating role of
prompts rather than the AI model itself, it sheds light on the ways journalists actively shape AI
infrastructures while negotiating public service values. The study contributes to debates on the
role of PSM in platformised public discourse and democratic values in times of digital capitalism.