They look human, are supposed to behave like humans and interact with them. They are in fact photorealistic computer generated images, animated with artificial intelligence technologies to generate behaviors as true as nature. A Samsung laboratory began to unveil, Tuesday January 7 at CES in Las Vegas – annual meeting of electronics and new technologies – avatars called “Neons”. This is'“virtual beings created on computer” who have, according to the communication of the South Korean giant, “the ability to show emotions and intelligence”.
Pending a real public demonstration on the part of Star Labs, Californian subsidiary of Samsung, some have already expressed their skepticism. Like the specialized site The Verge (in English), who writes: “Neons” are “just human-like digital avatars, which deserve the nickname of artificial humans as much as Siri”, Apple's voice assistant. “It is not a surprise, there are several laboratories in the world” who work to produce avatars animated by artificial intelligence, comments in turn Laurent Chrétien, the general manager of Laval Virtual, the most important French professional fair of virtual reality.
When the Neon project is truly revealed to the public, Laurent Chrétien will therefore seek to assess not the graphic quality of the avatars, but their ability to react to changes in their environment. “The real issue is the association of artificial intelligence and virtual reality: will we be able to create new universes, with virtual beings who have their own faculties?” if he asks.
Star Labs seems convinced in any case to have hit hard: “Neon is a new form of life”, don't hesitate to affirm Pranav Mistry, the boss of the lab. Core R3, the technological platform that will animate avatars, “can now independently create new expressions, new movements and dialogues (even in Hindi), completely different from the data originally recorded”, he tweeted on Saturday.
Flying to CES tomorrow, and the code is finally working 🙂 Ready to demo CORE R3. It can now autonomously create new expressions, new movements, new dialog (even in Hindi), completely different from the original captured data. pic.twitter.com/EPAJJrLyjd
– Pranav Mistry (@pranavmistry) January 5, 2020
But these avatars “could also be used as false human beings seeking to deceive a person for malicious or illegal purposes”, comments for independent analyst Jack Gold for AFP. The announcement could raise concerns, while social networks are fighting against the proliferation of misinformation and “deep fakes”, these hyperrealistic video montages designed to amuse or deceive the public.