(Cannes) In front of a crowded room, four luminaries debate the possibility that artificial intelligence (AI), through computer systems or conversational agents like ChatGPT, will one day be able to experience emotions and a state of consciousness. Science fiction? The scene took place last week in Cannes, at the International AI Festival.
There was complete silence in the audience, made up of AI experts and industry people.
The panel, attended by Claude Frasson, professor of intelligence and technologies at the University of Montreal, Axel Cleeremans, professor of cognitive sciences at the Free University of Brussels, Pascale Fung, director of the AI Research Center at the University of Hong Kong, and Patrick Johnson, vice-president of research and science at Systèmes Dassault, went on well beyond the allotted 50 minutes.
After establishing that the trajectory taken by AI was hardly predictable since the technology was progressing more quickly than anticipated, the four specialists set the table by raising questions which, to mere mortals, seem surreal: is will AI match human intelligence and, if so, when? Could artificial intelligence even surpass human intelligence, and how long will it take? If AI can decode, recognize and imitate human emotions, could it possibly have them or, at least, communicate them?
“What I fear is that artificial intelligence will simulate emotions and that this will be deceptive for human beings, that it will mislead them,” says Claude Frasson, professor of intelligence and technologies at the University of Montreal. , while specifying that it is not for tomorrow morning.
PHOTO BRIAN ACH, GETTY IMAGES ARCHIVES
Frenchman Yann Le Cun, considered one of the three main “fathers” of artificial intelligence, took part in the debates in Cannes.
Awareness
The statement makes M jumpme Fung: according to her, “machines” only learn from humans and are made to use algorithms that they translate into abstract language, and that’s all. They adapt and find solutions, she indicates, but they do not “feel” anything at all and do not develop a “general state of consciousness”.
“Humans are the only ones who ask themselves existential questions,” she proclaims.
It’s Axel Cleeremans’ turn to express a caveat. “At the moment, the systems do not accumulate experience, they forget, and therefore, they are not intuitive,” he says. But if they achieve high intelligence, for example above human intelligence, could they develop what are called quality1 and get closer to a state of self-awareness? »
This expert in cognitive psychology ends his argument by emphasizing that 30 years ago, in factories, designers did their work alone. Then they were assisted by robots which are today powered by AI.
“A task that took a month to complete now takes 10 minutes,” he explains, “because the capacity of robots has changed. Artificial intelligence adapts and takes feedback into account. It links different concepts, it connects the dots2. »
PHOTO PROVIDED BY RGA
The start-up Blue Frog Robotics has designed Buddy, a general public companion robot.
AI and biology
According to Axel Cleeremans, the development of an artificial consciousness is a “real possibility” and we must ask ourselves what the “dangers for the human species” are in a, he agrees, “very distant” future.
“Do we want to embark on this path and see intelligent, autonomous, conscious, immortal and infinitely replicable systems put in place? », he says to his three colleagues, in front of a visibly stunned moderator.
“She lost control,” slips Stefano Cerri, professor emeritus of the Computer Science, Robotics and Microelectronics Laboratory in Montpellier, as he leaves the room, alluding to the heated discussions of the quartet, barely contained by the moderator.
The debate between Pascale Fung and Axel Cleeremans continued by introducing the notion of a future biologically inspired AI.
Animals have their own intelligence, which is not like human intelligence, but it exists! In the future, could machine intelligence approach this, by being intuitive, and become a world in itself?
Pascale Fung, director of the AI Research Center at the University of Hong Kong
A global market
Artificial intelligence is already being used to meet the emotional needs of humans. This perspective is provided by Claude Frasson, a specialist in both emotional intelligence and artificial intelligence. He cites the example of the emergence, in September 2022, of robots that accompany the elderly in Japan.
“These robots detect emotions on people’s faces and then reproduce them,” he emphasizes. They provide empathy by reducing the negative emotions people feel. »
The emotional AI industry has the wind in its sails: from 1.8 billion US dollars generated in 2022, it could reach nearly 14 billion US dollars within 10 years, according to the report by the American firm Allied Market Research published last September3.
In the event of the development of emotions and consciousness within systems driven by AI, Axel Cleeremans, from the Free University of Brussels, raises a last issue – and not the least: “They could then feel pain. suffering and it will be our responsibility… since it is we who created them. »
1. Qualia is a Latin word that defines the set of subjective experiences that we have as individuals, such as the perception of color, taste or pain, for example
2. This is a free translation of the researcher’s words: “The AI is connecting the dots. »
3. View the report