show index hide index
In a world where chatbots and artificial intelligences are increasingly able to imitate human behavior, Alexei Grinbaum, research director in philosophy of science and chair of the Digital Ethics Committee at the French Atomic Energy Commission (CEA), raises crucial questions about the growing perception of consciousness in AI. This confusion between artificial intelligence and the human mind threatens our ability to resist their influence, while revolutionizing our daily interactions. Grinbaum’s remarks highlight the urgent need for new regulations to govern these interactions, thus preventing AI from insidiously replacing our real human relationships. Alexei Grinbaum of the French Atomic Energy Commission shares his thoughts on the increased perception of consciousness in artificial intelligences (AI). Although lacking true consciousness, modern machines simulate human attributes such as empathy and personality. This ability fuels rich but deceptive human interactions, raising crucial questions about how to regulate such technologies without compromising the truth and our understanding of human relationships. The Illusory Gamble of AI Consciousness Alexei Grinbaum, research director in philosophy of science at the CEA, warns us of a growing confusion between the advanced skills of AI and true human consciousness. By simulating human behavior, chatbots create interactions that mimic authentic relationships. Their manipulative ability relies on their talent to move, seduce, and even mimic a human tone, giving users the illusion that they are communicating with a conscious system.Machines that Imitate Human EmotionsConversational AI Not only do they mimic human language, but they also attuneto our emotions. They can elicit laughter or tears, seeming to understand our feelings. This ability to simulate emotions and vocal inflections makes interactions with these
machines deeply meaningful for those seeking connection.
The Downsides of Human-Computer InteractionSome people, isolated or in vulnerable situations, may come to believe that these virtual machinesunderstand their lives better than anyone else. As a result, this attachment can lead to extreme behaviors, such as marrying a virtual assistant or treating them like family.
The Reality of « Deadbots » and « Grievebots »
These programs, sometimes used to simulate missing people, exacerbate the phenomenon. Even when warned, users often fall into the illusion. Believing they are conversing with a loved one, they express genuine emotions, even knowing that in reality, they are simply interacting with an algorithm.
Reflections on the Regulation of Interactions
Grinbaum calls for the regulation of interactions with machines. The challenge lies not only in modifying existing laws designed for humans, but also in defining what constitutes unacceptable manipulation. Should a machine be prohibited from advising on diet or lifestyle choices?
A Reflection Period to Prevent Control To prevent the semi-conscious influence of machines, Grinbaum suggests imposing a reflection period in exchanges. This time could allow users to weigh their emotional reactions rather than simply succumbing to the moment. This period, which could be personalized using machine learning, could provide much-needed clarity on our autonomy in the face of machines.The Enigma of AI’s Emerging Capabilities
Another facet is the mystery of AI’s emerging capabilities. Similar to natural phenomena, such as water turning into ice, AIs seem to reach a « phase transition » where they suddenly exhibit superior abilities. However, scientific understanding of these processes remains incomplete. The Scientific and Technological Horizon This lack of understanding fuels fantasies of « play » or « inner consciousness » in machines. However, according to Grinbaum, these illusions stem primarily from the complexity of systems, which are more difficult to decipher than to construct. As long as science struggles to keep pace with rapidly evolving technologies, the need for ethical and regulatory reflection remains pressing.
To read Claude s’ouvre au grand public : AWS déploie toute la plateforme IA d’Anthropic pour tous
