Have you ever thanked ChatGPT? To say “please” to Claude? This can make you smile. We could see it as nothing more than a simple polite expression in front of a helpful and eloquent interlocutor. Some will say that it is better to remain civil, in case the machines one day reveal that they were conscious all along and want revenge!
The tone is light, but it denotes a very real phenomenon: through fluid, empathetic and personalized responses, conversational agents can give the feeling that they understand our thoughts, our emotions, or even that they are endowed with a form of consciousness.
As more and more people turn to chatbots for advice, comfort or companionship, this confusion can have real consequences. In an article published inThe Transmitter,a team of neuroscientists from the University of Montreal and Johns Hopkins University, in the United States, recalls an essential distinction: intelligence should not be confused with consciousness.
The team, made up of Vanessa Hadid, postdoctoral fellow in the Department of Psychology at UdeM and the McGill University Health Center, Karim Jerbi, professor in the Department of Psychology at UdeM and researcher at Mila, and JohnW. Krakauer, director of the Center for Restorative Neurotechnology at Johns Hopkins University, argues that a system can behave intelligently and react eloquently to our emotions without truly understanding them, without caring about us and without have any experience of feelings whatsoever.
For the authors, the more convincing and present these agents become in our lives, the more necessary it is to clarify that intelligent behavior, even fluid, reassuring or emotionally adapted, is not proof of consciousness.