Researchers at Korea University have developed a dual-output artificial synapse to boost the energy efficiency of multitasking AI systems, the university announced. The device emits both electrical and optical signals simultaneously to enable parallel processing. Tests showed up to 47 percent faster computation and energy use reduced by as much as 32 times compared to conventional GPU hardware.
Researchers at Korea University's KU-KIST Graduate School of Converging Science and Technology, led by professors Wang Gun-uk and Park Young-ran, have created a brain-inspired artificial synapse. Conventional AI chips are designed for specific functions, often requiring sequential processing for multiple tasks, which boosts power use. This new device allows parallel handling of tasks on a single chip by emitting electrical and optical signals at the same time.
The team reported stable learning across about 1,000 distinct states. In tests, it achieved up to 47 percent improvement in computational speed and reduced energy consumption by as much as 32 times versus GPU-based accelerators.
"This achievement presents a new hardware architecture for multitasking AI through an artificial synapse that simultaneously utilizes electrical and optical signals," Wang said. "It could be further expanded to high-speed, low-power AI systems in fields requiring complex decision-making, such as robotics, medical and health care applications, and autonomous driving."
The study appeared Friday in Science Advances, a journal from the American Association for the Advancement of Science. As AI demands more computing power and electricity, efforts to mimic the brain's efficiency are gaining traction.
