Readers consider a story composed by a computer algorithm just as credible as one written by a live journalist, demonstrates a paper by Hille van der Kaa and Emiel Krahmer, both of Tillburg University. The result has attracted interest and surprise after it was presented earlier today at the European communication conference in Lisbon, despite having already been published a few weeks ago at the Computation + Journalism Symposium in New York.
According to the study, upon reading the sample stories professional journalists would prefer the piece written by a live colleague, while as lay public saw it equally convincing as the one created without human oversight. The results are especially interesting because, unlike in previous experiments, the participants were made aware of the stories’ origins, and yet the (lay) readers saw automated story creation on par with that of human production.
The conference paper detailing the experiment can be found here.