Chinese scientists unveil energy-efficient 'brain-inspired' AI chip

Chinese researchers have made a breakthrough in chip manufacturing technology with the development of Speck, a new low-power "brain-inspired" chip designed for artificial intelligence (AI) applications.

A collaborative effort led by the Chinese Academy of Sciences (CAS), Speck is a neuromorphic chip capable of dynamic computing. It combines a dynamic visual sensor and a neuromorphic chip on one single chip, achieving remarkably low resting power consumption. 

This translates to impressive energy efficiency as Speck consumes a mere 0.7 milliwatts when handling visual tasks, said Li Guoqi, a researcher at the Institute of Automation of CAS, adding that this makes Speck a promising solution for AI applications that prioritize low power consumption and responsiveness.

The chip's innovation lies in its system-level design, integrating algorithms, software and hardware. This approach leverages the advantages of "brain-inspired" computing, mimicking high-level brain functions like dynamic attention allocation. 

The research, recently published in Nature Communications, highlights a significant step towards replicating the human brain's remarkable efficiency.

"The human brain is a marvel of neural networking," Li explained, "consuming just 20 watts of power, a fraction of what current AI systems require."

With ever-increasing computational demands leading to higher energy consumption, Li emphasized the potential of neuromorphic chips inspired by the brain's structure and function. 

Human brains can dynamically allocate attention based on stimulus, a process known as the attention mechanism. This research proposes "neuromorphic dynamic computing," applying this principle to enhance neuromorphic chip designs, thereby unlocking greater performance and energy efficiency, said Li. 

The research paves the way for the development of intelligent computing systems that are not only powerful but also remarkably energy-efficient, Li added.