Send message to Authors

Short abstract:

Journalists' understanding and use of AI shapes the news we receive. As embedded practitioner-researchers, we reflect on an action research project exploring journalists' imaginaries of AI and their implications for news work and generated interventions to build critical capacity in the newsroom.

Long abstract:

Journalists are dealing with rapid changes in the media and communication technologies they use for work, whilst having to simultaneously make sense of these technologies for audiences. As such, their imaginaries of AI and use of AI shape the news our societies receive. While generative AI and deepfakes make headlines, the incremental and ad-hoc instances of AI- and data-driven automation are quietly shifting the conditions within which news is produced. In this paper, we discuss an embedded action research project at public service broadcaster, the BBC. It explored newsworkers' understandings of AI but also aimed to intervene and engender change in the newsroom. Journalists held diverse and often ambivalent perspectives. Many reflected prevalent narratives about how AI-automated work processes could improve productivity and efficiency, be objective, and free up time from repetitive tasks for creative/investigative work but also make them redundant, introduce errors/biases, or lack empathetic and contextual qualities. A common thread was viewing AI as everywhere and nowhere – an invisible and inscrutable presence within core digital infrastructures, but which they couldn't pinpoint. The authors – a journalist and a research scientist at the BBC, who are also academic researchers - used participatory methods to co-design plausible futures and visual artefacts with which to interrogate professional implications of specific applications. We reflect on the benefits and shortcomings of this combination of etic and emic knowledge and argue for approaches that close the gap between theory, critique, and practice by having ‘skin in the game’ and making normative underpinnings explicit.