Scientists Develop Ultra-Efficient AI Model Inspired by Brain

Scientists Develop Ultra-Efficient AI Model Inspired by Brain
—————————-
Researchers have created a highly compact artificial intelligence (AI) model that mimics the brain’s visual system, potentially transforming both neuroscience and AI applications, a study published in Nature reports. Unlike conventional AI systems, which consume massive computing power, the new model uses far fewer variables while performing nearly as well, offering insights into how human and primate brains process visual information so efficiently.

The model simulates V4 neurons, which encode colors, textures, and shapes, using data from macaque monkeys. By compressing a system originally using 60 million variables down to 10,000, scientists were able to observe how individual artificial neurons respond to specific visual features, such as curved shapes or small dots. The findings suggest that brains achieve high visual recognition with minimal energy through specialized, streamlined processing.

Experts say the research could improve understanding of diseases like Alzheimer’s and guide the development of AI systems that are smaller, faster, and more “human-like,” potentially allowing applications such as self-driving cars to operate with less computational power while maintaining accuracy.