Human Brain Processing Can Inspire Next-Gen AI Systems, Researchers Say

Research published on Jan. 22 in Nature suggests that human information processing can serve as a model for training next-generation AI systems.

The study brought together over a dozen researchers worldwide, including Cory Merkel, associate professor of computer engineering at Rochester Institute of Technology. Merkel specializes in neuromorphic computing, a brain-inspired approach aimed at enhancing processing power and energy efficiency in AI applications.

“The ability to have efficient AI on constrained devices will also open the door to many new application domains in areas like brain-computer interfacing, space exploration, health monitoring technologies, and autonomous surveillance systems, for example,” Merkel explained, in the university press release.

His work addresses the growing demand for AI systems tailored to size, weight, and power-constrained environments, such as wearable devices, smartphones, robots, drones, and satellites. Neuromorphic computing promises significant improvements in processing capabilities and mass storage needs.

The researchers highlight how neuromorphic systems leverage bio-intelligence principles identified by neuroscientists, offering a model for faster and more efficient computational networks.

Merkel and Suma George Cardwell, a senior researcher at Sandia National Laboratory, also explored emerging memory technologies, such as RRAM and Spintronics, for mass storage in neuromorphic systems. These technologies show potential for scalable solutions and effective handling of device variabilities.

As AI’s electricity consumption is projected to double by 2026, researchers view neuromorphic computing as a promising solution. They highlighted that the field is at a “critical juncture,” with scalability becoming a crucial measure of progress.

Neuromorphic computing presents a path toward creating more efficient, energy-conscious AI systems for the future.