MetaAIOpensNewFrontierswithBrain

作者：小小 · 2026-06-29 21:54:20

Meta AI has unveiled two groundbreaking studies demonstrating the ability to decode sentences directly from non-invasive brain signals, marking a potential turning point in assistive communication. The research successfully reconstructed up to 80% of typed sentences solely from magnetoencephalography (MEG) activity, without requiring surgical implants. This approach captures the precise timing of neural responses, allowing the AI to interpret the dynamic process of language generation rather than just static thoughts. In a parallel breakthrough, Meta’s team explained how the brain transforms abstract ideas into coherent sentences, revealing a dynamic neural code that sequences words and meanings in real time. While clinical application remains distant due to current hardware constraints, the findings provide a deeper understanding of human cognition and lay a robust foundation for non-invasive brain-computer interfaces that could one day restore voice to those with severe communication impairments.