New research involving scientists from the Hebrew University of Jerusalem, Princeton University, and Google Research suggests that human brains process spoken language in a step-by-step sequence remarkably similar to advanced Artificial Intelligence (AI) Large Language Models (LLMs). This discovery challenges traditional linguistic theories by proposing that meaning emerges gradually from context.

• The study, published in Nature Communications, found that early brain responses to speech correlate with shallow AI layers, while later activity, particularly in Broca’s area, aligns with deeper, more complex AI processing.
• Researchers from the Hebrew University of Jerusalem, Princeton University, and Google Research were involved.
• The team has released its full dataset of brain recordings and language features to support further study into how the brain deciphers natural speech.