新型人工神经元能模仿脑细胞电化学行为，有望降低能耗提升AI效率

Core Insights - A research team from the University of Southern California has developed a new type of artificial neuron that mimics the electrochemical behavior of biological brain cells, marking a breakthrough in neuromorphic computing technology [1][3] - This innovation is expected to significantly reduce chip size and energy consumption, facilitating the advancement of general artificial intelligence (AI) [1][4] Summary by Sections Neuromorphic Computing - The new artificial neuron integrates a diffusion memristor and a resistor stacked on a single transistor, resulting in a compact and energy-efficient design suitable for neuromorphic computing systems [3][4] - Each artificial neuron occupies an effective area of approximately 4μm², showcasing the potential for miniaturization in chip design [3] Mechanism of Action - Unlike traditional digital processors or silicon-based "brain-like chips," this artificial neuron truly replicates the operational mechanisms of biological neurons, utilizing chemical reactions to drive information processing [3][4] - The neuron operates through a physical process that mimics the brain's signaling, where electrical signals are converted into chemical signals at synapses, triggering subsequent electrical signals in neighboring neurons [3][4] Efficiency and Structure - The diffusion memristor used in this artificial neuron allows for high-fidelity reproduction of neural processes at the physical level, relying on ion diffusion and memory effects rather than electron flow [4] - This design significantly simplifies the structure, requiring only one transistor per artificial neuron compared to traditional designs that may need dozens to hundreds, indicating a potential reduction in chip size and energy consumption by several orders of magnitude [4]