US scientists at the University of Texas have developed a non-invasive AI-powered decoder that translates thoughts into text in real time. This breakthrough, detailed in the journal Nature, moves beyond previous brain-computer interfaces that required invasive surgery. The decoder analyzes functional magnetic resonance imaging (fMRI) scans to produce intelligible word sequences as subjects either listened to or imagined narratives. While not a precise transcription, the device captures the gist of thoughts, achieving a 50% accuracy in reflecting intended meanings. Testing involved three participants exposed to 16 hours of stories, and the decoder leveraged a language model similar to GPT-1. Researchers emphasize ethics, ensuring the technology is only operated by willingly trained individuals and addressing potential misuse concerns. This advancement holds promise for restoring speech in patients who have lost their communication abilities due to injury or disease, marking significant progress in non-invasive brain-machine interaction.
Scientists Develop AI Decoder Capable of Translating Thoughts into Text
