Artificial intelligence, the Facebook friend of tomorrow?

Perhaps soon you will ask Paris Hilton’s avatar to suggest a cooking recipe. Or for your favorite YouTuber to correct their messages, before updating their schedule for the week. This is the unlikely announcement that electrified the great annual mass of the company Meta, at the end of September 2023. These “virtual assistants” will no longer be satisfied, as GPT Chat, to answer any questions in writing. They will also have a voice, an animated face (possibly inspired by a celebrity of your choice, therefore) and even their own personality (funny, serious, clumsy, etc.). They must be gradually put into service in Facebook, Instagram, WhatsApp and Messenger. The first were launched in December in the United States.

“These assistants are the future!” They will serve as intermediaries for all our interactions in the digital world.”supports Yann Le Cun, director of artificial intelligence (AI) research at Meta. This Frenchman also praised his merits during an Arcom study day a few months ago in Paris, insisting on the fact that these chatbots (conversational agents) are “much smarter than us” why “guardians of all human culture”. These so-called “generative” AI tools actually create never-before-seen content after having “grinded” huge masses of data taken from the Internet.

First step towards the metaverse?

For Olivier Ertzscheid, professor-researcher in information sciences at the University of Nantes, these Meta announcements are linked to the fiasco of metaversea ghostly project that became brutally obsolete with the success of ChatGPT in late 2022. “The company clearly sees that it will not be able to sell us “completely virtual” immediately. These customizable chatbots are a first step towards the 3D social world that Meta wants to see happen. »

This expert, author of a book on Facebook in 2020 (1), however, remembers that the digital assistant that “knows you better than you know yourself” It’s old fashion technology. The first half of the 2010s also constituted the first golden age of these technologies. But neither Siri (Apple), nor Alexa (Amazon), nor Assistant (Google) were based on “to transform”, the neural network architecture that underpins ChatGPT – and makes it so powerful. During last September’s developer conference, Meta CEO Mark Zuckerberg gave some details about these specimens. Your objective will not only be “answer questions” ; they will also have to “distract”. “We have arrived here in an uncanny valley: are we going to develop relationships with these technologies? “, asks Olivier Ertzscheid.

Mirror relationship

Certainly, according to recent tests. In April 2023, the social network Snapchat launched its MyAI chatbot. In two months, 150 million internet users had already “discussed” fashion, football or car with that “friend” virtual, presented as being more “fun” than ChatGPT. “Nowadays, social media algorithms select the content they present to us based on their ability to please us and capture our attention”recalls Matthieu Peltier, professor of philosophy at the EPHEC business school in Brussels. “It’s a safe bet that the configuration of these conversational AIs will be based on the same principle: meeting our expectations. » In other words, these “assistants” are just a mirror of ourselves and always tell us what we want to hear.

What this philosophy columnist for the Belgian radio station Première fears is that getting used to these artificial “relationships” will decrease our tolerance towards “other true” : one that disturbs our certainties. “When we see what digital technology has already done to our capacity for attention, our capacity for relationships does not seem out of danger to me”, highlights Matthieu Peltier. Some of your students already tell you that they can no longer watch an entire film – too long, too boring.

(1) The world according to ZuckerbergC&FÉd., 2020, €15, 112 p.

Leave a Comment