How readers perceive computer-generated news articles

A recent study investigates how readers perceive computer-generated news articles.

The advent of new technologies has always spurred questions about changes in journalism – how it is produced and consumed. A recent development which has come to the fore in the digital world is software-generated content. A paper recently published in Journalism Practice investigates how readers perceive automatically produced news articles vs. articles which have been written by a journalist.

The study, undertaken by Christer Clerwall of Karlstad University in Sweden, was conducted by presenting readers with different articles written by either journalists or computers. The readers were then asked to answer questions about how they perceived each article – e.g. the overall quality, credibility, objectivity.

The results suggest that the journalist-authored content was observed to be coherent, well-written and pleasant to read. However, while the computer generated content was perceived as descriptive and boring, it was also considered to be objective and trustworthy. Overall readers found it difficult to tell which articles had been written by journalists, and which were software-generated.

Perhaps most significant in Clerwall's study is the discovery that there were no substantial differences in how the different articles were perceived by readers. Does that mean that computer robots are capable of doing as good a job as journalists? Should journalists be considering a career change just yet? There are certainly advantages to be had in the speed with which computer-generated content can be produced, but will a robot writer ever be able to match the creativity, flexibility and analysis of journalist authored articles? The technology in place may not be quite able to reach these levels of sophisticated reporting yet, but it certainly provides food for thought as to how automated content might influence journalism in the future.