Brain-Inspired AI Model Learns While Shrinking

Researchers have developed a new framework for Spiking Neural Networks (SNNs) that mimics human infant brain development ^[1][2]. The model strengthens cross-regional connections while selectively removing redundant local links ^[1][3]. This biological approach allows the AI to handle complex tasks in perception and motor control ^[2]. This method significantly reduces the model's size and energy footprint ^[1][3]. Unlike traditional compute-heavy systems, this strategy supports continual learning across multiple domains without losing previous skills ^[2]. The discovery represents a shift toward sustainable AI inspired by the brain's natural efficiency ^[1].