Limits of robot journalism do not stop its advance
A new study shows that robots are not good journalists - but because they are inexpensive, there is still a great need for them.
According to a research paper published on Thursday, journalists and editors are convinced that so-called "robo-journalists" do not have a good feel for news and can only write one-dimensional stories. However, the report also shows that despite these limitations of robot journalism, there are plans to develop the technology further. Thomson Reuters alone is considering replacing "hundreds" of journalists.
The report, published in the international peer-reviewed journal Digital Journalism, presents the results of a survey of journalists, editors and executives from CNN, BBC, Thomson Reuters, Trinity Mirror and News UK, led by Prof. Neil Thurman (LMU Munich and City, University of London). All had hands-on experience with automated text creation software from one of the leading technology providers. The term robot journalism, also known as robo journalism, refers to software that creates text from structured data* with little or no human assistance. Such software is already being used by news services such as the Associated Press, the Los Angeles Times and Forbes.
Articles created by robots lack creativity...
The journalists and editors in Prof. Thurman's study believe that robot journalism's reliance on data streams and the need to define angles in advance mean that articles lack background, context, complexity and creativity compared to traditional reporting. One BBC journalist said: "I would never have written an article like this", while a veteran sports journalist from CNN judged the result to be "repetitive and just good for the bin". He felt that much of what he thought should be in an article "can't be programmed in".
The journalists also agreed that the fact that a template has to be created beforehand for automatically written texts is a major disadvantage. One BBC journalist, after his own first hands-on experience with the technology, said: "You can't question anyone or get explanations about the figures because the template for the article was already written before the figures came in. In his opinion, the technology was not worth the BBC's time.
...but the desire to reduce costs is great
But despite these weaknesses, journalists believe that robotic journalism can help reduce costs and improve speed and accuracy in certain types of reporting. Journalists from CNN and Reuters felt that robot journalism could "reduce costs" by replacing "expensive staff" who do "relatively simple and time-consuming work". A Reuters journalist said that automation could improve the speed and accuracy of reporting. This could be "interesting for all areas of the company". Another Reuters journalist said that automation would be used for articles "that cannot be covered manually due to a lack of staff".
What about artificial objectivity?
Robot journalism was seen as both a means of promoting and a threat to journalistic objectivity. One Sun journalist said that robot journalism could "present facts as they are" without "manipulation" by humans. This could be "quite a good tool for democracy". However, a BBC journalist feared that automated journalism and the volume it could generate could make it easier for "partisan" individuals and organizations to influence news priorities.
Prof. Neil Thurman concludes: "Robotic journalism is limited in its ability to provide the current context needed for reporting, to understand the nuances of human expression that determine how an event is reported, and to always recognize the most important angle. However, we think that this form of journalism will be used more in the future for simple factual reporting, for faster publication of such reports and for topics that were previously not worth reporting on."
"However, the huge increase in news volume due to automated text creation can make it harder to navigate a world already saturated with information, and even increase the need for the human skills of good journalists - news judgment, curiosity and skepticism - to keep us fully, accurately and succinctly informed about the world around us."
*For example, Opta's English Premier League data feeds with details of up to 2000 events, from the score to the number of goals and missed shots.
You can download the detailed version of the study here download (PDF in English, direct link).
