For a Human-Centered AI

Artificial journalism. Issues that are not that new

December 11, 2023

The great enemy of journalism is not Artificial Intelligence. There is worse, for example, the uncritical use of its know-how, falsehood and lack of accuracy.

On March 2, FBK Magazine hosted an editorial of mine, devoted to the possible impacts of artificial intelligence on journalistic content in the future. The cue came from reading the interviews that the Nieman Foundation offers every year-end to female journalists to think about the horizons of their profession.

The future is (also) now, and the relevance of AI on journalism is (also) in the news. A few days ago, here’s what set me thinking, the media company Futurism published an article in which it exposed a bad practice of Sports Illustrated.

The facts. Articles written by a fictitious Drew Ortiz rang alarm bells for their neglected tone. Ortiz was described very generically as an outdoorsman and nature lover, an assiduous camper and hiker. He seemed like the ideal profile for those tasked with recommending useful tools for living those passions, but some doubts arose: no biography of the dashing Drew, no profile on social networks, and, on top of all that, a photograph that scratching the surface turned out to be generated by artificial intelligence. Delving a little deeper, editors at Futurism realized that not only Drew Ortiz’s profile had been produced by artificial intelligence (as were those of Sora Tanaka, Somino Abrams, and others), but also his articles. Pressed, those at Sports Illustrated blamed the source: the pieces had been outsourced to a third-party agency (AdVon Commerce), which allegedly ensured they had been written by human hand (and mind). According to anonymous allegations inside Sports Illustrated, this sounds suspicious, to say the least; in fact, it is widely believed that several articles are actually products of artificial intelligence. Even Maggie Harrison, the journalist who signed off on the Futurism investigation, is skeptical: “The implication seems to be that AdVon invented fake writers, assigned them fake biographies and AI-generated headshots, and then stopped right there, only publishing content written by old-fashioned humans. Maybe that’s true, but we doubt it.”  Implausible it seems, in short, that we put so much effort into building virtual journalists and then have real “colleagues” write their pieces.

Opinion.   What is wrong with the choice of Sports Illustrated, or AdVon, depending on who one might consider to be the main protagonist in this reliance on Artificial Intelligence? There are two elements to be mainly taken into consideration.

The first one. The domestication of truth, provided it is not a lie. Pretending that people write things written by machines is deceptive: readers (who pay for many services) need to be clearly informed about the authors of what is offered to them. If Drew Ortiz or Sora Tanaka are deliberately created pleasing young pens, they should know it.  If the perspective from which they compile work is that of an artificial intelligence and not a human one, one must know it. In order to be appropriately informed, those who read must be appropriately informed

The second one. Sloppiness, not that of the AI, but of those who use it uncritically. Harrison noted that the pieces attributed to invented identities had not been reviewed and contained passages that were difficult to attribute to human error (not reviewing articles by human journalists is also a problem, of course) because they reported real nonsense, for example, that you need a ball to play volleyball. A re-reading would have resolved the inaccuracy. I don’t think it is in itself wrong to ask artificial intelligence for help, but you can’t do it without any exchange with it.

The future is now, but it seems intent on retracing paths journalism has known for centuries: falsehood and approximation. Knowing the problem could help find a solution to it, provided we want to.


The author/s