While invasive neural implants like Neuralink capture headlines by drilling micro-electrode arrays directly into the cerebral cortex, non-invasive brain-computer interfaces (BCIs) have quietly undergone an AI-driven revolution.
Historically, non-invasive EEG and optical functional near-infrared spectroscopy (fNIRS) were plagued by the skull’s severe signal attenuation and low spatial resolution. Today, self-supervised spatial-temporal transformers pretrained on tens of thousands of hours of neural signal recordings can reconstruct semantic linguistic intent and motor cursor trajectories without requiring invasive neurosurgery.
