Pierre Bessiere will discuss four key points made in his latest book (released nov. 20, here):
- The first message is that for an intelligence, having a “body” changes everything. Natural intelligences, from Chlamy, a single-celled algae, to humans, all have bodies. The most powerful artificial intelligences do not. Robots have bodies but are still very rudimentary.
- For artificial intelligences, in the absence of a body, “meaning” is only given by a human user. Without this human mediator, artificial intelligences are meaningless. They perform calculations, but these have no impact on their environment unless a human intervenes.
- For natural intelligences, having a body allows them to interact directly with their environment. This interaction, and the resulting actions and perceptions, is what gives “meaning” and allows us to predict what is likely to happen: if I perceive this, then I will do that and I will perceive that. On the other hand, this interaction with the physical world is so complex that it cannot be fully captured by any intelligence whatsoever. Intelligences with a body are irretrievably ignorant, and the Truth of the physical world eludes them.
- The Truth is inaccessible to them, but Reason, the art of making the best use of the little information available to them, is indispensable. The book therefore compares two models of reasoning: logic and probability. Logic, the current foundation of our computers, which deals only with Truth. Probability, which allows us to reason while being ignorant, with incomplete and uncertain information. Probability is indispensable for reasoning correctly while being ignorant.
Le livre étant pour l’instant en français l’exposé sera en français avec des transparents en anglais.
As the book is currently in French, the presentation will be in French with slides in English.
About Pierre Bessière
Pierre Bessière is an emeritus director at the CNRS, working within the ISIR laboratory at Sorbonne University. He led the Bayesian Programming research project, whose central focus is the probabilistic modelling of perception, inference, learning, and action. The group’s core thesis is that subjectivist (Bayesian) probabilities offer a simple mathematical theory of cognition and rationality—an alternative and an extension to logic.
Pierre Bessière has worked on these topics at the University of Grenoble, INRIA, the Stanford Research Institute, and at the Collège de France in Alain Berthoz’s laboratory, and now at Sorbonne University. He co-founded two companies based on this line of work: Probayes and HawAI.tech.
Details
Date and Time: Thursday, 27th of November 2025 - 2 PM
Location: Sony CSL, 6 rue Amyot, 75005 Paris
Registration: here