Premiere: ChatGPT drafts an amendment soon to be debated in France

For the very first time, a text partly written by ChatGPT will be debated in the National Assembly following an amendment tabled by the deputies of the Liot group (Freedoms, Independents, Overseas and Territories), a group that came out of the woodwork recently with a mention of cross-partisan censorship aimed at bringing down the government. If the vote failed on March 20 by nine votes, their amendment could (should?) go all the way.

The height of the height: artificial intelligence (AI) calls for more human controls in a bill relating to the Paris 2024 Olympic Games adopted in January by the Senate, aiming to use artificial intelligence for video surveillance systems .

It goes without saying for Acquaviva

Even if the deputy of the second constituency of Haute-Corse Jean-Félix Acquaviva, at the origin of the amendment in order to alert on the risks of this type of new technologies, recognized that the answer provided by ChatGPT has been "slightly retouched" to stick to French law, it must be admitted that the amendment proposed by ChatGPT (then in version 3.5), the AI of the US company OpenAI, around article 7 of the bill, flows from source. And for good reason. Article 7 authorizes the experimentation of biometric augmented cameras during sports competitions to analyze the flow of images in real time, in particular to identify crowd movements or suspicious behavior.

Never too careful?

Questioned by the Liot deputies, ChatGPT recommends adding the following statement to the text: "Subject to respect for the fundamental rights and freedoms of the persons concerned and the obligation to guarantee human decision-making at each stage of the processing" . Well said ChatGPT, right? Sometimes an old law on AI dating from 2023 allows algorithms to be given much more freedom in the future... Sources: Capital and La Lettre A