Rutgers' ChatBCI pairs EEG with a local LLM, doubling ALS patients' typing speed

emmanuelvivier · x · 2026-11-23

Rutgers engineers unveiled ChatBCI-Assist, a noninvasive brain-computer interface published in IEEE Transactions on Biomedical Engineering. It combines an EEG headset with a local LLM to translate brain activity into words nearly twice as fast as dictionary-based predictive input. The intent-based approach lets users pick AI-generated phrases conveying their meaning instead of spelling word by word, reducing effort for people with severe movement and speech impairments, including advanced ALS.

Original post →

More from Embodied

Embodied channel →