“I just landed in Marseille. The soil glue, the air stinking” Exclaims an alien by leaving his flying saucer. Because yes, thanks to artificial intelligence, extraterrestrials discover our dear Marseille! But as evidenced by these first words, the video with humorous purposes spreads the clichés in Marseille.
Dozens of videos like this circulate on social networks, particularly on Tiktok. The characters and the sets are very realistic, we recognize the Old Port very well, the Julien course but also the Vélodrome stadium.
“My wife told me that I had to choose between her and my football match, I told her I needed 90 minutes to think” Exclaims an elderly gentleman, laughing, the scarf of Olympique de Marseille around the neck, surrounded by other supporters.
A tool used to convey shots
At the origin of these videos, the new tool generating images of Google “I see 3” which allows “to add sound effects, ambient noises and even dialogues to your creations, generating all sounds in a native way “, Indicates Google on its site. Launched in the United States in May 2025, it has only been used in Europe since early July. To create a video, nothing could be simpler. It is enough to describe in a few lines the scene that we want to see, and the AI reproduces it.
If some videos are good, most degrade the image of the city and its inhabitants, while conveying racist ideas. On one of them, we see a young woman with sunglasses walking in the streets of Marseille while filming, like a vlog. She documents her “stay in Marseille.”
“Day 3 in Marseille I am in the northern neighborhoods. I have just stolen my watch, but it does not matter, the holidays continue“She said with a big smile. Behind her, the street is full of waste. Then, she continues,”Day 5 in Marseille, I was assaulted only twice this morning. ” Over the days, she wears the hijab then the burka, and ends up speaking Arabic.
Racist videos
The racist bias is confirmed by the description of the video. The Internet user wrote: “Process of integration of an influencer in Marseille. Seriously, we will have to identify these people.” “But who of the user or the machine is at the origin of the racist character of the content diffused?” questions TV5 world.
In a publication, a user writes that when he asked the AI “a banal experience in Africa”, she created a white man, selfie stick in the right hand and bottle of water in her left hand, who is followed by a black child asking for water. “A scene that perpetuates the image of a poor Africa requiring the help of white men”Comments the TV5 world journalist.
This is not the first time that an artificial intelligence tool has been pointed out for problematic content. In 2016, Microsoft and Bing released Tay, an AI for conversational purposes. Less than 24 hours after its creation, the tool made racist, hateful responses and sticking false information.
