ALBUQUERQUE, NEW MEXICO – Physicists at Sandia National Laboratories have reported a major leap in their work to steer LED light, crediting a trio of artificial‑intelligence systems for accelerating results that previously took years of experimentation.
The team originally announced the ability to steer LED light in 2023, a discovery with potential applications in scanners, holographic projectors, autonomous vehicles, and other technologies that currently rely on lasers. Researchers expected the refinement process to be slow, but a newly published study shows their AI‑assisted approach improved performance fourfold in about five hours.
According to Sandia scientists, the “self‑driving lab” setup uses three different AI tools. One system learns and simplifies complex optical data, another designs and runs experiments on its own, and a third generates equations that explain what the AI is discovering. This approach not only produced rapid improvements but also helped researchers interpret why those improvements work—an important requirement for scientific progress.
The findings, published in Nature Communications, suggest that LEDs could someday replace lasers in a range of technologies thanks to cheaper, smaller, and more energy‑efficient components. The discovery also represents a broader shift toward AI‑driven scientific research, where machine learning tools can test hypotheses, refine designs, and generate explanations at speeds traditional lab work cannot match.
