Does artificial intelligence (AI) have a negative or positive influence on journalism? Difficult to answer such a question, as its ramifications are multiple and complex. What is certain is that AI is profoundly changing the media sphere with increasingly powerful tools. “ While AI will facilitate the production of misleading content and make it harder to detect, it is also a tool to detect the manipulation of information and facilitate journalistic work ” said Noël Corbin, General Delegate for Transmission, Territories and Cultural Democracy at the Ministry of Culture. Friday, March 22, actors from the media came to address, as part of the Digital Culture Meetings the impact of this technological revolution in their work and the initiatives undertaken to educate their public about AI.
AI is a great tool, but…
Without a doubt, AI offers tremendous economies of scale. At Radio France, it has been used since 2016 for very practical uses of real-time program transcription. Without a doubt, AI offers tremendous economies of scale. “ This is a positive use of artificial intelligence because previously, the journalist had to transcribe everything by hand, which was long and tedious "explains Estelle Cognacq, deputy director of Franceinfo, in charge of digital and director of the Radio France agency. Also very useful, tools for detecting unpredictable news: thanks to some tweets detected by the AI, some newsrooms were immediately aware of events such as the attacks in Strasbourg or Belgium.
But everyone knows that AI makes it possible, with a disconcerting ease, to format false information almost undetectable on social networks. But you don’t stay unarmed to protect yourself. “ I believe in the collaboration between research and journalists in the fight against disinformation, says Estelle Cognacq. AI and its important development require us to go there and fight on equal terms. » Radio France has partnered with Inria’s engineering school to develop a statistical tool capable of verifying information in real time.
Detecting information manipulation: from critical analysis to critical AI
Agence France Presse was also upset about a year ago with the arrival of generative AI accessible to the general public such as ChatGPT or Midjourney, especially in the field of image. It launched AFP Factuel, which operates in 85 countries with 140 journalists in 26 languages, which deals with dubious claims that circulate on the Internet and are potentially harmful. « We deal with topics of public interest that have a certain virality on social networks ” summarizes Alexis Orsini, a fact-checker journalist with the service. Sometimes with an ant’s work to trace the source of certain images produced with AI. “ The best weapon we have is to go back to the very first instance on the web to find its author. These are hyper manual methods, says Juliette Mansour, a «fact-checker» journalist with the AFP Factuel service who is optimistic about our ability to recognize the truth of the fake. We must appeal to his sense of analysis: many clues are hidden in the photos. »
But while analysis is often safer, it takes time. “ We make a mistake by considering AI as an enemy of democracy, explains journalist Thomas Huchon. The problem is the distribution of messages: anyone can say anything and social networks will make sure that the message reaches a huge audience. That’s why there’s nothing better than using AI itself to flush out false information and intervene against it as virally as possible. Thus was born last January Anti fake news AI ”, materialized by an avatar of the journalist. When false information is detected, Huchon can produce a contradictory video using artificial intelligence very quickly – in less than an hour. The character we see on the screen is not real but « words are spoken by humans, as well as investigations, writing and editing ” he says. This method makes it possible to respond to this time issue, which is crucial to bring a more powerful contradiction on social networks and for the public to form a contrasting opinion.
Reporters Without Borders: addressing ethical, independence and legal issues
Reporters Without Borders took up the subject by creating, in June 2022, on the principle of the Munich Charter, the European reference for the ethics of journalism, a Paris Charter for AI. “ The challenge was to bring together experts with different sensitivities to reach a consensus on common values to defend to preserve the specificities of journalists "explains Vincent Berthier, head of the RSF technology desk. This document is based on 10 principles related to the production, collection and dissemination of information.
RSF also launched the Spinoza project, with the Alliance for General Information Press as its first partner, which brings together a dozen press groups. This experimental approach aims to manufacture in-house an artificial intelligence tool that operates on the data that editors make available to it, so that they free themselves from AI that does not belong to them. Finally, RSF has launched a third project on deep fake – or video infox – with the creation of an international offence for using this process.
Partager la page