“Do a newspaper only with the Artificial Intelligence (AI)? Who I do, I wish you good luck. Will need it, “says Sarcastic Claudio Cerasadirector of the Italian newspaper The sheet. Carasa knows what he’s talking about. It is what he did days ago: put on sale for a month, from Tuesday to Friday and Together with the paper newspaper Always, too an edition produced entirely by theat a time when the media discuss precisely how to use this technology.
The first conclusions of the initiative of The ia sheet They have revealed some obviousness derived from the intrinsic nature of AI and its current level of development. One of the most significant: The AI does not write items and offers new information, rewritten (or reelabate) content from news that human journalists have previously written. And, for the moment, with the added and disturbing risk of the disinformation.
So far from the project, another of the great findings is that, in addition, the AI commits inaccuracies and errorsin some cases also serious. “The platform to which we have asked to write the articles of The ia sheet es Chatgpt for [la versión avanzada, de pago y mejor que la gratuita] y Always require human supervision and review. The reason? An example: it resists accepting that Donald Trump He is the current president of the United States, “he says, during a presentation of the initiative in the writing of the newspaper in which the newspaper participated.
Claudio Ceras, director of ‘Il Foglio’, in the newspaper of the newspaper in Rome. / Irene Savio
So much so that, If the human supervisor does not know properly the argument of the article that must produce the AI, The risks multiply. The explanation is in the interaction itself between the journalist and the AI that, in the case of the experiment of The sheet, Start with the journalist asking AI to write an article through a question that he writes on the platform.
“If the question is not formulated in a correct way, if it is too generic or the content that the article must include and relate, the risk of the AI will be wrong or cheat is not required, and that is a constant challenge. Even when the editorial line is detailed, the result is not always expected,” says Nicola Contarini, one of the reporters in charge of the project of this newspaper of this newspaper of the liberal line and conservative 1996. “When we were talking about The possibility that AI replaces To journalists, that gave me some scary. But, since I started working with AI, I feel much quieter; This fear I went quite fast“He adds.
Disinforms
Walter Quattrociocchi, data scientist And an expert in AI and algorithms, do not go with rodeos by describing as “delusional” just thinking that AI can be used to write newspaper articles. This is because, of course, AI either has ethics nor know the deontology, with The consequence inevitable collateral of the disinformationthe expert emphasizes.
The reason is that AI can “generate false information, simplify complex issues and produce contents that seem credible Even when they are not, “says Quattrociocchi.” But, above all, he can insert texts automatically generated in a journalistic format with the risk of legitimizing the idea that information can be produced Without responsibility“The analyst explains.
Therefore, in his opinion, the threat is in the industry. “Manyespecially among editors, They are not sure how they work These models and believe that you can really replace journalistic work, “he emphasizes. However,”Generative models do not know ‘anything: predict the following word based on past texts […] And it is not clear how they select the contents, “he adds, emphasizing the (other capital) problem of the fuentes that the AI choose for its elucubrations.
The writing of ‘Il Foglio’ in Rome. / Irene Savio
Light years
Who controls that it is reliable sources? Not to mention Right to freedom of expression or the journalistic style. Cerasa, the director finally admits it. “What we are seeing is that the quality of the information produced by The AI is light years of the of our journaliststhat do a much better job, “confesses the journalist and former blogger, adding that momentarily the most serious mistakes are being pointed out and that only when the project ends, on April 11, a more complete compendium of all the mistakes will be made.
Cerasa adds that still his writing has positively welcomed the idea and does not regret having created The ia sheet. “Our goal was to speak fearless of this topic. We need to start take the reins of this debate y Let’s not be governed by technologies. I think we are successful in that. They are writing and speaking of us around the world, “he says, remembering that his diary is also a small reality (22 journalists) that boasts of their uninhibited and provocative ideas.
