We have created the first quantum-optical associative memory, published in Science 394, 212 (2026)! We experimentally show that one can store and recall memories in a spin glass, which was long thought not possible. The exponentially many glassy minima—which cannot be used as memories under equilibrium recall—can now be turned into reliable memories under driven-dissipative cavity dynamics. By doing so, we achieve a storage capacity that exceeds the classical Hopfield limit by as much as sevenfold. A second key discovery is a polaronic “elasticity” that mimics the short-term synaptic plasticity inherent to biological systems. We show that this doubles memory capacity and points to future directions in which our neural network can be trained using naturally occurring synaptic plasticity. Read the paper here.
Highlighted “In Science Journals.”