Chinese scientists have developed a new brain-inspired artificial intelligence model that could help reduce the high energy demands of traditional AI systems. The research, published by the Institute of Automation under the Chinese Academy of Sciences, introduces an innovative approach that focuses on the internal dynamics of neurons rather than simply increasing the size of neural networks. Current AI models often rely on building larger and more complex neural networks, known as “external complexity,” which requires significant amounts of energy and computational power.
However, the human brain, with its 100 billion neurons and 1,000 trillion synaptic connections, can operate efficiently using only about 20 watts of power. Inspired by the brain’s efficiency, the Chinese researchers developed a network model that incorporates “internal complexity” within individual neurons. This approach allows the AI system to manage complex tasks while maintaining a smaller overall network size.
Brain-inspired neuron dynamics boost efficiency
Researcher Li Guoqi explained that the experiments conducted by the team, which included collaborators from Tsinghua University and Peking University, confirmed the effectiveness of the internally complex model in handling sophisticated tasks. The study proposes a new method for integrating neuroscience principles into AI development to optimize performance.
The brain-inspired AI model marks a significant step forward in the pursuit of more energy-efficient and interpretable artificial intelligence systems. By focusing on the internal dynamics of neurons, scientists aim to create AI that can operate more effectively while consuming less power, similar to the way the human brain functions. As AI continues to advance and find applications in various fields, the development of more efficient and brain-like models could help address concerns about the technology’s energy consumption and computational requirements.
The Chinese researchers’ work contributes to the ongoing effort to create AI systems that are not only powerful but also more sustainable and easier to understand.
