Love ChatGPT and Alexa? This year’s Physics Nobel Prize has been awarded to two scientists who made these AI technologies possible

In a world where artificial intelligence powers everything from our phones to our smart homes, it's easy to forget the humble origins of these powerful technologies. Today, the Nobel Committee reminds us of these roots by awarding the 2024 Nobel Prize in Physics to two extraordinary scientists — John Hopfield and Geoffrey Hinton — for their groundbreaking work on artificial neural networks, the cornerstone of today’s AI.

The story of neural networks

Artificial neural networks, the technology that drives modern AI, mimic the way human brains process information. These networks consist of nodes, or “neurons,” that connect and influence one another, much like synapses in our own brains. This network architecture allows AI to learn and improve over time, making it capable of tasks like recognising speech, translating languages, and even generating human-like text.

The Nobel laureates were instrumental in building this foundation, each contributing uniquely to the development of neural networks. John Hopfield, a professor at Princeton University, introduced the Hopfield network in the 1980s. His method allowed AI to recognise and reconstruct patterns, enabling it to “remember” data in a way that was previously unimaginable.
On the other side of the Atlantic, Geoffrey Hinton, a professor at the University of Toronto, was working on his own innovations. Building on Hopfield’s ideas, Hinton developed the Boltzmann machine, a system that could autonomously find properties in data and recognise patterns. This network introduced the possibility of teaching AI to classify objects and even generate new examples, and it remains a fundamental part of AI technology today.

Why physics?

It may seem surprising that the Nobel Prize in Physics, traditionally awarded for work in the physical sciences, went to two scientists whose work is more closely associated with computer science and AI. However, as the Nobel Committee explains, these advances are grounded firmly in physics. Both Hopfield and Hinton applied principles of statistical physics to train neural networks, using concepts originally developed to describe the interactions between particles.

The Hopfield network, for example, is based on the physics of atomic spin — a property that makes each atom a tiny magnet. Hopfields work utilised this concept to help his network “remember” patterns by aligning the connections between nodes to reduce energy, allowing it to retrieve images from memory in a manner similar to how a magnet aligns with a magnetic field.
Hinton’s Boltzmann machine, meanwhile, leverages statistical physics to enable machines to learn. The system trains by running repeated examples through the network, gradually improving its accuracy at recognising specific features — much like how we recognise familiar faces in a crowd.

Transforming the world of AI

Hopfield and Hinton’s discoveries laid the groundwork for many of today’s most powerful AI tools, from language models like ChatGPT to image-recognition software used in medical diagnostics and autonomous vehicles. Without these early innovations, the AI systems that have become so integral to modern life would not exist.

Their discoveries have far-reaching applications, from scientific research to the everyday technology we often take for granted. As Ellen Moons, Chair of the Nobel Committee for Physics, remarked, “The laureates’ work has already been of the greatest benefit. In physics, we use artificial neural networks in a vast range of areas, such as developing new materials with specific properties.”

The recognition of Hopfield and Hinton’s work is not only a nod to the past but a beacon pointing toward the future. As AI continues to evolve, the foundation they built will remain crucial, inspiring new generations of scientists and researchers to push the boundaries of technology. It’s a testament to the enduring impact of their work that we are only beginning to scratch the surface of what’s possible with AI.