Monografico_3_EN_Doxa_29

solution to summarize news instantly. In the same way, data scientists in Latin America are working on the first robot-journalist to report on legislative bills (Monnerat, 2018). One of the pioneers in Europe was The Guardian, which in 2010 was already working with two sports news automation projects to obtain game statistics as well as historical information about teams and players. In addition, the daily newspaper combined this data with pre-set phrases and connectors to compose stories (Bunz, 2010).

A year later, it launched an app that automated the search for current news through Twitter (González, 2011), and in 2014 it did the same with Guarbot, a program that writes financial information using complex data, preventing journalists from doing the job (Gani; Haddou, 2014). That same year, the news company even published a free monthly newspaper in paper format with an American edition, and then British. The selection of topics was done by AI systems. The BBC News Lab already works with SALCO (Semi-Automated Local Content) software to cover local news, and The Telegraph has developed its own bot known as Roboblogger, which allows it to publish an indeterminate number of views from the data generated at each sporting event that is covered live on the web. Moreover, the UK Press Association has developed RADAR, a system that allows local stories to be written for the media in an automated way thanks to the information they collect from open data sources in governmental departments as well as from regional and local authorities.

In Germany, Der spiegel, Neue Osnabrücker Zeitung, Weser-kurier, Radio Hamburg Fussifreunde, and Fupa.net (formerly Fubanews), also use automation to write sports news, as well as Handelsblatt press, which uses it for economic texts. The Berliner Morgenpost also uses AI to report on particulate pollution in Berlin.

In France, Le Monde used AI to generate articles during the March 2015 elections (Sánchez; Sánchez, 2017), and the Swedish daily Svenska Dagbladet uses it to create customised home pages of its website from an algorithm that utilizes only two datum provided by the journalist: a “news” value (between 1 and 5), and a longevity parameter.

The rest is done by the machine, which places each news item according to various parameters, such as clicks on different news, length of stay, or preferences shown by each reader in previous visits (Stern, 2017). Similarly, several agencies admit transmitting content generated by web robots, but they do so without specifying the subject matter.

These include the following: DPA (Germany); ANP (Netherlands); STT (Finland); AFP (France); APA (Austria); Ritzau (Denmark); Lusa (Portugal); NTB (Norway); and TT (Sweden).

In Spain, some delegations of Agencia Efe also work with small automated data processing systems (Fanta, 2017), as well as El País, which in collaboration with Google applies AI to managing the comments of its digital readers in order to raise the quality of discussion and encourage conversation on the newspaper’s platforms. This media also created a bot on Facebook Messenger to provide information about the French presidential elections in 2017. This was in addition to the one it already had for general news, with the latter having more than 10,000 subscribers (Southern, 2017).

In 2017, Vocento MediaLab launched the “Medusa” project (InfoPlayas and InfoEsquí), which experiments with journalism-robot models to automatically generate content from some 800 Spanish beaches and all the ski resorts in Spain, Andorra and the French Pyrenees. El Confidencial has created Ana Futbot, a bot developed by El Confidencial’s laboratory that is already publishing automatic chronicles of Second Division B soccer matches. At the same time, Politibot stands out as well. This program was born as a Telegram bot to cover the Spanish Presidential elections on June 26, 2016.