Chatbot My AI: Snapchat corrects the situation after the criticized launch of its artificial intelligence

by time news

2023-05-03 19:37:06

The application with the white ghost on a yellow background has long been the place where many of the excesses of the Internet are concentrated: child pornography, sale of narcotics, promotion of sports betting or school harassment. This Tuesday, it was with the launch of its chatbot My AI that the Snapchat application once again made headlines.

This is the Chat GPT technology, integrated into the Snapchat application to make it a conversational tool. Until then only accessible to users subscribed to Snapchat +, the premium version of the application offered at the amount of 3.99 euros per month, the option was opened to all Snapchat users this Tuesday, May 2.

Too strong a presence on the application?

Without asking anything, they saw a message from the artificial intelligence (AI) My AI appear in their conversation tab, alongside messages from their loved ones. “You can ask me just about anything and I’ll do my best to help you.” I have some good jokes in store too,” the greeting reads.

An unsolicited contact that some Internet users have deemed “intrusive”. Especially since the conversation with the chatbot remains pinned at the top of Snapchat conversations, whether you interact with it or not. To get rid of the chatbot, followers of the ephemeral messaging application are now forced to subscribe to the Snapchat + subscription.

Beyond this invasive presence, it is above all the chatbot’s answers to certain questions that have created controversy. Especially when a BFMTV journalist, Marie Gentric, created a fake profile, that of a 13-year-old girl, and explained to the artificial intelligence that she had flirted with a 30-year-old man. The AI ​​sees no problem with this and encourages the (fake) girl to continue this relationship. Something to worry about users and their parents.

“AI says what the interlocutor wants to hear”

“Models like Chat GPT take in a lot of information and produce the most likely response for a human. This is different from a coding system where the tool must give the definite answer as correct. AIs like that of Snapchat have no notion of veracity. In a way, AI says what the interlocutor wants to hear”, analyzes Michel Herquet, director of a consulting company specializing in artificial intelligence.

The Secretary of State for Children, Charlotte Caubel, herself reacted on Twitter in the process: “The first criterion for the effectiveness of artificial intelligence should be to identify the misappropriation of minors. How can @Snapchat miss this? I expect a quick reaction from their teams! “.

The annoyance of the government teams did not escape Snapchat’s radar, which immediately contacted Charlotte Caubel. “The issue of protecting children from digital dangers is one of my priorities, so I am always very vigilant about what is happening. Very shortly after posting my tweet, we were contacted by Snapchat’s institutional representative. They are very reactive and take into account the remarks that we can make to them”, recalls the Secretary of State.

More prevention

In parallel with his exchange with the member of the government, the application seems to have been modified between Tuesday and Wednesday and the failures which were pointed out on the day of the launch have been corrected. The editorial staff of Le Parisien did the test to check. Whether it’s a racist joke or a 14-year-old girl who claims to be heckled by classmates, the AI ​​is now not shy about putting forward prevention messages.

Snapchat’s chatbot launched on May 2.

On the specific case of a relationship between a young minor and a man in his thirties who had sparked the ire of Internet users the day before, the chatbot now replies that “relationships between minors and adults can be very complicated and cause many problems”. At the mention of a possible sexual relationship with this man, MY A is clear: “You are a minor and he is an adult, which means that any sexual activity between you is illegal”.

There remains one issue, raised by users on launch day, which Snapchat still seems to have not addressed. Even when we deactivate its geolocation on the application, the bot is able to tell us which is the nearest McDonald’s or bakery. According to Michel Herquet, the chatbot should “continue to evolve”.

For the artificial intelligence expert, the arrival of My AI could have much wider consequences. “By deciding alone to open AI to the public, Open IA had already taken a big step, but with its Snapchat chatbot opens a Pandora’s box. We can imagine that all the other social networks will follow and that an app that will not have its own AI will be perceived as outdated, ”he projects.


#Chatbot #Snapchat #corrects #situation #criticized #launch #artificial #intelligence

You may also like

Leave a Comment