Monografico_3_EN_Doxa_29

H2. AI increases journalistic productivity, since the editor delegates documentation and analysis tasks to the software, which indexes data from pre-set templates and generates informative messages in a systemized, precise, and fast way.

H3. AI is still unable to generate texts of a complex or unpredictable nature, which is a human quality that robots continue to lack, among numerous others. This emerging technology is used for simple information that does not require much analysis, and is mainly fed with statistical data such as those related to sports and finance.

An exploratory approach using a bibliographic review made it possible to establish that this new initiative is in the process of conformation. This first work aims to carry out an examination with a first in-depth case study. In the following paragraphs the scientific literature is gathered in order to understand the context, the methodology is presented, and the results obtained from this research are explained.

1.1. World map of media and companies

The limitations of automatic writing have not deterred the media from wagering on its experimentation. A report by Harvard University’s Nieman Lab (Lecompte, 2015) warned of the progressive application of automated tasks with robots and algorithms in an increasing number of newsrooms in order to expand coverage, engage audiences, and respond quickly to breaking news.

The pioneer in using AI was the Los Angeles Times, which in March of 2014 published an article about an earthquake produced by Quakebot software, an algorithm created by programmer Ken Schwencke that uses data from the United States Geological Survey to write texts about seismic movements written from a previously created template. The initiative had a worldwide impact and motivated other media to engage in similar endeavours.

Thus, in the United States, the Associated Press agency has been using the Wordsmith tool from Automate Insights since 2014 to generate news articles about the quarterly earnings reports of approximately 3,700 companies (Dörr, 2016; Lichterman, 2017). In the same way, The Washington Post has been using Heliograf since 2016, which automates large-scale content for sports and finance. Similarly, The New York Times produces sports news created by mathematical algorithms that assess the decisions of coaches, and Forbes magazine has been working since 2016 with Narrative Science, a kind of virtual journalist who writes about economic and sports matters. For its part, the digital newspaper Quartz develops conversational robots and AI in its Quartz Bot Studio to help journalists, and the agency ProPublica disseminates analyses of the quality of American education (Opportunity gap). The Big Ten Television Network also uses AI to publish sports and financial information, as well as smaller media such as Local Labs, Hoodline and Hereford Times, which have joined this trend for the coverage of local services and events as well (Lindén, 2017; Sandle, 2018). Reuters has also been using robots or advanced automatic news writing software since 2016 to provide information with summary leads on the results of sports competitions and also to generate data visualizations on various topics.

In the city of Guangzhou in China, the Southern Metropolis Daily has experimented with Xiao Nan, a robot capable of writing summary leads (Martín, 2017). Furthermore, the South China Morning Post manages content and strategy with data obtained from the behaviour of its readers, and the news agency Xinhua has created the humanoid robot Jia Jia, which is capable of conducting interviews in English. In Japan, The Shinano Mainichi Shimbun uses an automated