Est Républicain journalists tested the correction of their articles using Chat GPT

This is an unprecedented experiment at the editorial office of the newspaper L’Est Républicain, for three months. Journalists tested the use of artificial intelligence to correct articles.

For three months, journalists fromIs Republican lend themselves to unprecedented experimentation in the editorial office of the newspaper. They tested the use of artificial intelligence to proofread articles. Nearly 700 papers were reworked by Chat GPT, in its free or paid version. These are only articles written by local correspondents, who do not have the status of journalists but of independent workers. They cover the news of a municipality, whether it is a marriage, the opening of a business or a sporting event.

Usually, it is the editorial secretaries who correct, rephrase or choose a title. This winter, around ten of them agreed to entrust these tasks to artificial intelligence, while supervising the robot’s work. Result: a significant time saving, several minutes per text. On this point, management and unions speak with one voice. They also agree on the errors made by the conversational tool, with altered city names and modified quotes. Sometimes, the AI ​​tends to draw its own conclusions, or even give a little moral lesson.

This experiment sparked an outcry in November, to the point that the management of the Ebra group, to which it belongs The Republican East, had suspended its project at the end of last October. The project was finally restored a few days later and the test took place, as recounted Rue 89 Lyon.

Points of divergence between leaders and trade union organizations

For unions, AI responses are sometimes unsuitable. They also always offer the same writing style, which results in standardized texts. Management, for its part, believes on the contrary that the robot is rather reliable and that it allows the editorial secretary to concentrate on something else. The results are considered positive, in any case promising, so much so that the Ebra group, to which theIs Republican, wants to launch new experiments. A charter to regulate the use of artificial intelligence is also being established.

This is what other press groups have already done, such as Le Monde, Le Figaro, Les Echos-Le Parisien, Ouest-France. All have established a strict framework for the use of AI, specifying, for example, that it should not replace editorial teams. Chat GPT or others should only be a support tool, but one that the media doesn’t want to do without, it’s still fantastic progress. It helps with translation, transcribing an oral interview, making an audio version of written articles, spotting manipulated content, such as deepfakes, or false information in speeches by public figures. Always, obviously, under human supervision.