Summary and Outlook
Automated journalism currently works well in producing routine news stories for repetitive topics, for which clean, accurate, and structured data are available. In such situations, algorithms are able to generate news faster, at a larger scale, and with fewer errors than human journalists. Furthermore, algorithms can use the same data to tell stories from different angles, in multiple languages, and personalized to the needs and preferences of the individual reader. Also, software providers have started to release tools that allow users to automatically create stories from their own data.
Automated journalism cannot be used for domains where no data are available and is challenging where data quality is poor. Furthermore, algorithms derive insights from data by applying predefined rules and statistical methods (e.g., identifying outliers and correlations) but cannot explain new phenomena or establish causality. That is, while algorithms can describe what is happening, they cannot provide interpretations of why things are happening. Algorithms are thus limited in their ability to observe society and fulfill journalistic tasks such as orientation and public opinion formation.
Automation will likely change the way journalists work, although the extent to which technology will replace or complement journalists will depend on the task and skills of the journalist. In the future, human and automated journalism will likely become closely integrated and form a “man-machine marriage.” Journalists are best advised to focus on tasks that algorithms cannot perform, such as in-depth analyses, interviews with key people, and investigative reporting. While automation will probably replace journalists who merely cover routine topics, the technology is also generating new jobs within the process of developing news-generating algorithms.
The widespread adoption will ultimately depend on whether news consumers like reading the content. Evidence available to date—which is limited to topics where automation technology is already being used on a large scale (e.g., sports and finance)—shows that while people rate automated news as slightly more credible than human-written news, they do not particularly enjoy reading it since the writing is perceived as rather boring and dry (see Textbox 1). Therefore, the technology is currently most suited for topics where (a) providing facts in a quick and efficient way is more important than sophisticated narration (e.g., financial news) or (b) news did not previously exist so consumers have low expectations regarding writing quality. That said, the writing quality of automated news is likely to improve, as natural language generation technology advances further.
Other important questions for the use of automated journalism in newsrooms relate to issues of algorithmic transparency and accountability. In particular, little is known about whether news consumers (need or want to) understand how algorithms work, or about which information they use to generate content. Furthermore, apart from some basic guidelines and principles that should be followed when using automation technology, there’s little data about which information news organizations should make transparent and how their algorithms work (e.g., decision rules or underlying data). Such information may become particularly relevant in situations where (a) errors occur and (b) content is personalized to the needs and preferences of the individual news consumer. Finally, a potential increase in personalized news is likely to reemphasize prior concerns regarding filter bubbles or fragmentation of public opinion.
Automated journalism has arrived and is likely here to stay. The key drivers are an ever-increasing availability of structured data, as well as news organizations’ aim to cut costs while at the same time increasing the quantity of news. This guide summarized the status quo of automated journalism, discussed key questions and potential implications of its adoption, and pointed out avenues for future research. In particular, conducting future research into questions about how automation will change journalists’ roles and required skills, how news organizations and consumers should and will deal with issues relating to algorithmic transparency and accountability, and how a widespread use of automated and personalized content will affect public opinion formation in a democratic society would be valuable. Furthermore, that research should track how the writing quality of automated news evolves over time. In particular, it might consider how people’s expectations toward and perceptions of such content change—especially for controversial and critical topics, such as election campaign coverage, which are not merely fact-based and involve uncertainty.