“Today, there is no real inclusion of human cognition (…) We mustn’t hesitate to challenge the human being and its adaptability.” Célestin Sedogbo _______________ “The more we work on artificial intelligence, the more we become aware of the specificity of humans and their brain, which feeds on change and depreciates with routine.” Pierre-Marie Lledo
Artificial intelligence, cognitive science, behavioural psychology, neuroscience… So many topics covered during the Orange Summer research days, held last July. Guests of the “Neurosciences and Cognitive sciences” conference, Célestin Sedogbo, director of the Institut Cognition, and Pierre-Marie Lledo, director of the Institut Pasteur neurosciences department, emphasised the adaptability of the human brain in the age of the emergence of AI.
From regulation to automatism
Célestin Sedogbo, director of the Institut Cognition and language processing expert, presented a vision of intelligence in cognitive science, successively through the limits of the human-system interaction, the augmentation of humans by technological systems, and the challenges of cognitive technology.
“Today we have human-regulated automatic systems such as the flight management system (FMS) for pilots, and we are moving towards more and more autonomous systems that we call self-organised. From regulation to autonomy, the idea has always been to reduce human activity, based on the assumption that human beings are not entirely predictable or reliable. But at the level of internet services, human intervention is still substantial like for search engines and especially for email. Based on the notion of cognitive cost, in other words the propensity of the brain to reduce its energy consumption, the ideal would be for all technological innovation to be designed to minimise this cognitive cost, for example by optimising access to information. A practical case of taking the user benefit into account would be for example, to show the nearest service stations to car drivers only when their car’s tank is nearly empty. This isn’t the case. Today, there is no real inclusion of human cognition. Understanding cognitive effects in order to make calculable models enables us to design systems that are cognitive themselves, or orthoses that simplify our lives, making certain tasks easier and even increasing our abilities. Yet at the same time, we mustn’t hesitate to challenge the human being by offering it new even more complex tasks so as to mobilise the adaptability of the brain. An over-simplified world would inevitably lead to an impoverishment of our abilities. Cognition is a strategic challenge for French industry.”
(Also listen to: podcast: “Interfaces that are more and more intuitive”)
The 21st century brain: digital or social?
Pierre-Marie Lledo, director of the Institut Pasteur neurosciences department and of the CNRS “Genes, Synapses and Cognition” laboratory, has worked on defining the characteristics of humans compared with those of artificial intelligence, and on identifying the place of man side by side with machines.
“Tomorrow, will we all be uberised? Are algorithms going to anticipate or direct our decisions? In fact, the more we work on artificial intelligence, the more we become aware of the specificity of humans and their brain, which feeds on change and depreciates with routine. The human being is perfectible and grows from what surrounds it. The brain is dynamic, it is an ongoing construction site. After outsourcing of the skeleton with the spear-thrower invented by Neanderthals, then outsourcing of muscles with the steam engine, since the 1950s we have been living cognitive outsourcing with artificial intelligence. But with the expansion of the digital ecosystem and the multiplication of smart objects, how will we be able to process so much information? Yet innovation cannot be brought about by automatic processes. The added value of the human being is to provide decisions that are fed by emotional valence. In fact, it is in moments of introspection that humans are the most creative. “I find when I am no longer searching”, said mathematician Raymond Poincaré. What differentiates us from machines is the social dimension of our brain.”
At the end of this conference, Nicolas Demassieux, Orange’s director of research, wished to emphasize, above and beyond big announcements and declarations, the limits of transhumanism and the as yet modest achievements of the artificial intelligence community. “We tend to overrate the individual performance of such or such algorithmic module, at the expense of the human being’s social intelligence and especially at that of the networking of artificial and natural intelligences, which is much more promising in terms of innovation.”