The Algorithmic Gut Feeling: Epistemologies of data in AI-driven News Distribution

This article explores the epistemic practices and doxa of data workers in a news organisation in Denmark that is currently developing and experimenting with artificial intelligence (AI)-driven recommender systems, machine learning and natural language processing solutions. Previous literature on the changing epistemologies of digital journalism has focused on the increased role of metrics and the transformed practices inside newsrooms, as well as on how journalists perceive and articulate the computational. This article advances these studies by focusing on how data scientists perceive and articulate “the journalistic” when building AI systems for distributing news. Developing the notion of “the algorithmic gut feeling”, the article highlights different frictions present in the articulations of the journalistic doxa in AI-driven data work concerning (1) how to algorithmically define ethics, (2) how to algorithmically categorise and understand relevance, and (3) how to algorithmically curate “a good mix” for the front page. The emerging frictions and algorithmic gut feeling are key to understanding how the doxa of data workers involved and deeply invested in “the good of journalism” at times also transforms journalistic epistemologies of what constitutes “news” and “the right mix” of content in the service of a democratic public.