The 2024 Nobel Prize in Physics has been awarded to John J. Hopfield (Physicist) and Geoffrey E. Hinton (Computer Scientist) for, "foundational discoveries and inventions that enable learning with artificial neural networks".
You can find the full scientific background the Nobel Committee cited for the prize here. What has me excited about this one is that it represents a recognition of the fundamental connection between Computer Science and Physics.
This is my over-simplified, but hopefully reasonably accurate, take on this year's prize...
Physics constantly stumbles on situations where complex dynamic
systems find their own point of equilibrium based on the current set of
inputs. The first time I understood this was when I learned how a
carburetor balances fuel and air based on dynamic inputs.
This is essentially what this year's Nobel prize in Physics is all about.
Professor
Hopfield demonstrated that the "self leveling" behavior of dynamic
systems found in nature could be used to create physics inspired models that exhibited
learning behaviors like storing and reconstructing data.
As the Noble Prize background explains, "Optimization occurs through integration of the
equations of motion of an electronic circuit, during which the nodes evolve without instructions from a central unit".
Professor
Hinton extended this idea by demonstrating that a mathematical tool
called backpropagation could be used to create networks that can perform complex tasks like learning, pattern recognition, and prediction. What you understand as ChatGPT is really just a series
of neural nets that are good at predicting what the next word should be
when generating a sentence in response to your questions.