Agentic AI and journalistic knowledge making in PSM newsrooms - who orchestrates whom.

Reitzer, PaulineORCID: https://orcid.org/0009-0009-3819-6436 Agentic AI and journalistic knowledge making in PSM newsrooms - who orchestrates whom. 8th Weizenbaum conference, “Generative AI and Society: What is at stake”, June 10-11, 2026, Berlin, Germany.

[thumbnail of Weizenbaum_Conference.pdf] Slideshow
Weizenbaum_Conference.pdf - Presentation
Restricted to repository staff only

Download (3MB)

Abstract

European PSM newsrooms rejected AI as a research tool, citing concerns about bias, hallucinations, and the use of probabilistic systems as risky for journalistic practice. Yet, within the short period of the last year, AI “agents” for research purposes were becoming central to PSM newsroom innovation, reflecting a broader transformation in how journalistic knowledge work is organised. This rapid adoption responds to mounting challenges: the proliferation of synthetic content, the persistence of fake news, and the sheer volume of information now available make traditional research increasingly complex and time-consuming. In this context, agentic AI emerges as a logical solution, capable of navigating vast, fragmented, and machine-readable data landscapes. These developments situate journalism within what could be termed the era of intra-activity, in which machines do not merely assist humans, but increasingly communicate and operate with other machines, reshaping knowledge-making in ways that carry direct implications for democratic public life.

In my project I examine how communicative AI reshapes journalistic practices when newsrooms introduce “AI agents” as research tools. As journalists will increasingly collaborate with agentic AI in the research phase of news production, the paper asks: How does agentic AI affect knowledge-making processes in the PSM newsroom?
The paper draws on ethnographic observations from an ongoing case study at a European public service media organisation and follows the introduction of an AI research agent throughout its “beta” phase. Methodologically, the AI tool itself is approached as an ethnographic field site. The analysis combines an examination of its technical affordances over time (prompt-based walk-through), as well as it explores its “black box” through semi-structured execution logs. These materials are analysed with regard to the communicative patterns through which the system structures and anticipates journalistic knowledge-making practices.

Preliminary findings connect transforming journalistic research practices to broader transformations in the ecosystem of information retrieval and synthesis and underline the need for further research on agentic systems in knowledge-making processes in the news industry and across domains.

Item Type: Conference or Workshop Item (Paper)
Event Organiser: Weizenbaum Institute
Research Units: Digitalization and social transformation
Former Research Units: Former Research Groups (until 2024) > Science, Technology and Social Transformation
Date Deposited: 01 Sep 2026 13:38
Last Modified: 01 Sep 2026 13:38
URI: https://irihs.ihs.ac.at/id/eprint/7504
  • View Item