Reimagining machine vision with optical computing

Researchers developed a general-purpose artificial intelligence vision system by embedding core computer vision operations into an optical metasurface on June 17, 2026. This light-manipulating planar material enables real-time perception and processing for diverse tasks, offering a potential solution for rapid, low-energy, on-device vision intelligence. The prototype demonstrates accurate performance across various image-sensing applications, suggesting a significant advancement in machine vision capabilities. This optical computing approach bypasses traditional electronic processing, potentially leading to substantial reductions in power consumption and increases in processing speed for AI vision systems. The findings, published in Nature, highlight the transformative potential of integrating computational functions directly into optical materials for future AI hardware.