Using AI to decode language from the brain and advance our understanding of human communication
Source check: Source checked. Not independent scientific confirmation.
Finding: A deep-learning decoder translated non-invasive EEG and MEG signals recorded during typing into sentence characters in an offline demonstration with 35 healthy volunteers. Mean character error rates were 67% for EEG and 32% for MEG, with a best MEG result of 19%. Participants typed memorized sentences; the work does not establish unrestricted or real-time communication.
Technology: A deep-learning decoder maps non-invasive EEG/MEG signals recorded during typing to sentence characters.
Publication: Announcement only · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- A deep-learning decoder maps non-invasive EEG/MEG signals recorded during typing to sentence characters.
- Test and result
- The primary preprint evaluated 35 healthy volunteers; mean character error rates were 67% for EEG and 32% for MEG, with best MEG 19%.
- Limitation / still unknown
- Healthy skilled typists, memorized text, offline decoding and shielded-room MEG limit generalization; distinct from invasive handwriting BCI.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://ai.meta.com/blog/brain-ai-research-human-communication · https://arxiv.org/abs/2502.09874v1
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: blog. Paper: linked separately.
Review date: 22 Nov 2026. Review scope: Official Meta announcement and linked primary Brain-to-Text preprint abstract and full text; first preprint version checked..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 4/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.