The Automated Newsroom in Australia and beyond: Problems and challenges in the use of automated decision-making systems in journalistic practice
Focus Area: News and Media
Research Program: People
This interview-driven research project provides a comprehensive overview of current journalistic designs and values following the implementation of automated tools and Artificial Intelligence (AI) in Australian commercial and public service newsrooms. It delves into the current configurations of editorial automation, specifically during the pandemic and before the development of generative AI times in datafied societies. The study explored how journalists, editors, and developers imbue with journalistic ideals the creation and industrialisation of automation processes, data-centric tools and types of AI in different stages of editorial work ranging from information gathering, data journalism, automated writing and summarising, producing and editing, visualising, distributing and recommending.
Drawing on a purposeful sampling method, that is, choosing practitioners who were most likely to be working on a daily basis with automated technologies, 17 journalists from the leading national and regional outlets, were interviewed out of 100 professionals invited to participate during the pandemic years. The study concludes that while this critical incident impelled journalists to be more familiar with data journalism practices and data management processes, the situation was not enough to accelerate complex automated processes such as, for example, automated text generation or any ML uses. The amount of COVID-19-related data being monitored, collected and then delivered to newsrooms by government and public entities (e.g., cases, infections, deaths, hospitalisations, and numbers of vaccinated Australians, among other structured data) animated some journalists and developers in Australia to incorporate, adapt and reassess to a certain extent some data tasks embedded in the production routines during this acute incident. However, automated-generated text during the pandemic was not sustainable in the long term due to the characteristics of the Australian news market, and audience news avoidance. Some participants reported that these automating processes were challenging to sustain also because of the evolving consumption trends during the pandemic. On the other hand, the study also outlines increasing uses of AI commercial tools, such as transcription applications and some advances in implementing news recommender systems (NRS) and voice assistants, such as the ABC emergency voice assistance.
Commercial outlets have progressively adopted algorithmic systems regarding news recommendations, but the development “is still in its infancy”. Although the study could not cover the popularisation of generative AI, unveils trends and perceptions of AI, which would require an expansion in a second round of interviews.
The second stage of this project replicates the Australian study in Latin American newsrooms (with a project agreement titled: Envisioning automated journalism in Latin America). Currently, collaborators on this project from La Sabana and the lead investigator are interviewing outlets practitioners in the region. To undertake the Latin American study, the lead investigator submitted an ethics extension based on Project 421 to be able to interview participants in that region until 2024.
The project developed one important ramification and collaboration with the Centre-affiliated investigators and one Latin American University. With the ADM+S affiliated researcher, Dr Michelle Riedlinger, from the Global Journalism Innovation Lab (GJIL) and the Digital Media Research Centre, and Dr Victor Garcia Perdomo from Universidad de La Sabana (Colombia), as well as Dr Marina Joubert from Stellenbosch University, we submitted a project to the Research Foundational Integrity Meta Grant. This study was awarded funding in December 2022 and started in 2023. It incorporates questions on automated fact-checking in counteracting misinformation and disinformation. The overarching investigation delves into the uses of social media algorithmic tactics deployed by Meta-affiliated fact-checkers in the Southern Hemisphere to counteract COVID-19 vaccine problematic content. These organisations follow the Meta fact-checking policies and have access to automated and AI platform tools for misinformation detection. We explored their uses and algorithmic practices with a particular focus on the Southern Hemisphere.