Researchers have developed a new artificial intelligence system that can translate brain activity into written text without requiring surgically implanted electrodes. The technology combines advanced deep learning with noninvasive brain recordings, offering a safer path toward helping people who have lost the ability to speak or type due to paralysis or neurological disorders. The system, called Brain2Qwerty, was trained using brain activity from 35 healthy volunteers as they memorized and typed sentences. It analyzed signals captured through electroencephalography (EEG) and magnetoencephalography (MEG), learning to predict the characters participants intended to type based solely on their brain activity. The best results came from MEG recordings, with the AI achieving an error rate as low as 18% for some participants and accurately decoding sentences it had not previously encountered. While the technology is still in the research stage, it significantly narrows the performance gap between invasive and noninvasive brain-computer interfaces. Researchers believe the breakthrough could pave the way for safer communication technologies that enable people with severe motor impairments to communicate, control digital devices, and potentially operate assistive technologies without the need for brain surgery.
