The Nobel Physics committee surprised everyone this year. By celebrating, on Tuesday, October 8, two pioneers of “artificial neural networks”the American John Hopfield (91 years old) and the British Geoffrey Hinton (76 years old), follows the current trend of artificial intelligence, which we would more easily associate with computing.
“It is the recognition that a current of physics, statistical physics, has made the effort to advance towards other fields. This is good news”says Rémi Monasson, CNRS researcher at the ENS Physics Laboratory in Paris. Stéphane Mallat, professor at the Collège de France, greets an award “amazing” and points out that, in return, artificial intelligence helps physicists a lot today, to obtain images, modeling, simulations, etc.
It’s not easy to see physics concepts behind the words written by ChatGPT, the images created by Midjourney, the videos generated by Sora or AlphaGo’s brilliant Go moves. It also doesn’t help that one of the two winners, Geoffrey Hinton, is not a physicist, but a computer scientist and neuroscientist. And still.
Transform a network into memory
The most talked about artificial intelligence systems at the moment belong to the category of machine learning, or “machine learning” in English, and more precisely to the subcategory that uses the mathematical model of artificial neural networks, a digital set of active or inactive neurons, linked more or less strongly to each other. John Hopfield, then at Caltech University (California), and Geofrey Hinton, at Carnegie Mellon University (Pennsylvania), in the 1980s, independently demonstrated that this mathematical analogy with the brain could surprise and do things usually reserved for our organ: memorize, learn. , recognizing patterns… “It is the illustration of what, in our field, we call emergency: the whole is worth more than the sum of its parts”summarizes Marc Mézard, professor at the Bocconi University of Milan. Physicists had already demonstrated this power in their discipline. A simple set of needles placed with their heads up or down, side by side on a checkerboard, can represent the properties of a magnetic material. Giorgio Parisi, Nobel Prize winner in 2021, master in statistical physics, this science that describes macroscopic phenomena based on individual behaviors, had developed this philosophy for more complex materials.
You have 62.49% of this article left to read. The rest is reserved for subscribers.
