The Nobel Assembly at the Karolinska Institute in Sweden announced on Tuesday that the 2024 Nobel Prize in Physics goes to American scientists John Hopfield and British Geoffrey Hinton for their work on machine learning techniques by developing artificial neural networks and the underlying algorithms that allow machines to learn, which is what we mean today when we talk about artificial intelligence.
Hopfield is a professor at Princeton University in the United States, while Hinton is a professor at the University of Toronto in Canada. The projects that the Nobel laureates worked on are key to models of artificial intelligence, big data, and modern software such as ChatGPT.
Ellen Moons, a member of the Nobel Prize Committee at the Royal Swedish Academy of Sciences and chair of the Nobel Committee for Physics, said that the laureates “used fundamental concepts from statistical physics to design artificial neural networks that act as associative memory and find patterns in large data sets.”
She explained that such networks have been used to advance physics research and become part of our daily lives, such as facial recognition and language translation technologies.
While AI has not been a strong contender for the Nobel Prize in Physics, the discovery of neural networks that can learn and their applications are closely related to physics, Moons said: “These artificial neural networks have been used to advance research in various physics topics, such as particle physics, materials science, and astrophysics.” Applications of the technology include many early approaches to AI, such as providing computer programs with logical rules to help solve problems.
In 1982, Hopfield at Princeton University created a computer system called a Hopfield network, a collection of connected nodes, or artificial neurons, that can vary the strength of their connections using a learning algorithm that Hopfield invented. Hopfield’s algorithm was inspired by work in physics that finds the energy of a magnetic system by describing it as a collection of small magnets. The technique involves repeatedly varying the strength of the connections between magnets to find a minimum value for the system’s energy.
That same year, his British colleague Geoffrey Hinton at the University of Toronto was working on a closely related machine-learning architecture called a Boltzmann machine, a machine that could learn and extract patterns from large data sets, and which has led to the success of many of today’s AI systems, such as image recognition and translation tools.
The announcement came as one of the laureates, Hinton was making a public appearance warning about the dangers of the AI he helped develop. Hinton told the Nobel committee that he might regret the work he had done: “In the same circumstances, I would do it again, but I worry that the overall consequence of this might be that systems that are smarter than us eventually take over.” Hinton’s concerns led him to quit his job at Google so he could speak more freely about the risks of the technology he helped create.
