The Royal Swedish Academy of Sciences has jointly awarded the 2024 Nobel Prize in Physics to John J. Hopfield and Geoffrey E. Hinton, for research carried out in the 1980s that forms the algorithmic infrastructure of today's machine learning systems.
Source: nobelprize.org
In plain terms
The 2024 Nobel Prize in Physics honours two researchers who borrowed the tools of physics to make machines process information. John Hopfield drew on the behaviour of atomic spins: his network "settles" into its lowest-energy state, like a marble rolling to the bottom of a bowl, which lets it recover an image from a damaged version. Geoffrey Hinton added the element of chance found in statistical physics: his Boltzmann machine picks out, on its own, patterns that recur across large amounts of data. It is these two ideas from the 1980s, rather than any recent result, that form the first layer of what is now called deep learning.
The prize at a glance
| Parameter | Value |
|---|---|
| Award | 2024 Nobel Prize in Physics |
| Institution | Royal Swedish Academy of Sciences |
| Attribution | Joint |
| Laureates | John J. Hopfield and Geoffrey E. Hinton |
| Period of the work | 1980s |
| Grounds | Use of the laws of physics to model information processing |
| Hopfield's contribution | The Hopfield network, based on the physics of atomic spins |
| Hinton's contribution | The Boltzmann machine, based on probabilistic physics (statistical mechanics) |
| Reach | Algorithmic infrastructure of today's machine learning systems |
| Primary source | Official press release, NobelPrize.org |
The two contributions
1. The Hopfield network — memory as energy minimisation. John Hopfield designed a model, now known as the Hopfield network, based on the physics of atomic spins and intended for the associative reconstruction of data. By simulating energy minimisation in complex physical systems, these "neural networks" correct distortions in images.
2. The Boltzmann machine — probabilistic physics in the service of features. Building on that work, Geoffrey Hinton brought in probabilistic physics (statistical mechanics) to produce what is called the Boltzmann machine. This system makes it possible to identify autonomously isolated structural features within large volumes of data, for example characteristic elements in images, becoming the first layer leading to the discipline of "Deep Learning" as exploited by modern computing.
The two models side by side
| Criterion | Hopfield network | Boltzmann machine |
|---|---|---|
| Author | John J. Hopfield | Geoffrey E. Hinton |
| Physics drawn upon | Physics of atomic spins | Probabilistic physics (statistical mechanics) |
| Principle | Energy minimisation in complex physical systems | Autonomous identification of structural features |
| Demonstrated use | Associative reconstruction of data, correction of distortions in images | Spotting characteristic elements within large volumes of data, for example in images |
| Lineage | Starting point | Builds on Hopfield's work; first layer of "Deep Learning" |
Causal chain
Physics of atomic spins and statistical mechanics → the Hopfield model (1980s), simulating energy minimisation → associative reconstruction and correction of distortions in images → Hinton brings in probabilistic physics → the Boltzmann machine and autonomous identification of features → first layer of "Deep Learning" → algorithmic infrastructure of today's machine learning systems → the 2024 Nobel Prize in Physics, awarded jointly by the Royal Swedish Academy of Sciences.
Sources
References checked during the August 2026 fact-checking audit: these are the pages
against which the claims in this bulletin were verified.