AI Breakthrough Uses Light to Create Images With Minimal Power
TEHRAN (Tasnim) - Researchers have developed an AI system that generates images using light instead of heavy computer processing, cutting power consumption to nearly zero.
As the use of generative AI expands in fields ranging from art and coding to drug design, its growing demand for energy has raised global concerns.
In a paper published in Nature, Aydogan Ozcan of the University of California, Los Angeles, and his team introduced a light-based image generator that sharply reduces power consumption.
Traditional AI systems generate images using a diffusion process, where a dataset is trained with layers of statistical noise that are later reversed to form the requested image. While effective, this method is slow and energy intensive, especially when applied on a large scale.
The new system replaces heavy computer calculations with an optical process. It begins with a digital encoder creating static, which is then imprinted onto a laser beam through a spatial light modulator (SLM). A second SLM decodes the pattern, producing the final image.
“Our optical generative models can synthesize countless images with almost no computing power, offering a scalable and energy-efficient alternative to digital AI models,” said lead author Shiqi Chen.
The researchers tested their system on diverse images, including celebrity portraits, butterflies, and full-color artworks in the style of Vincent van Gogh. Results were found comparable to conventional AI systems but achieved with far less energy.
Experts say this technology could significantly cut the carbon footprint of AI content and pave the way for new uses in augmented reality, virtual reality, and compact devices such as smartphones and AI-powered glasses.
