Meta has introduced a revolutionary system, Brain2Qwerty v2, capable of reconstructing typed text from brain activity recordings without surgical intervention. The average word-level accuracy reached 61% — an impressive leap compared to roughly 8% for previous non-invasive methods.
How the technology works
Brain2Qwerty v2 was trained on a dataset of approximately 22,000 sentences. The experiment involved nine healthy volunteers, each spending 10 hours in a magnetoencephalography scanner, typing heard phrases on a keyboard. Instead of manually creating schemas to identify neural events, the developers applied end-to-end deep learning, decoding data directly from raw brain signals.
The system works with continuous brain activity recordings and reconstructs the entire sentence. A fine-tuned language model, accounting for semantic context, is used to refine the result. The best-performing participant achieved 78% word-level accuracy — more than half of their sentences were decoded with one error or none at all.
Current limitations
The development is still far from clinical or everyday use. Brain2Qwerty v2 was tested exclusively on healthy volunteers who actually typed text. For people who have lost speech or the ability to move, this approach requires separate validation. Additionally, the system does not function as real-time text input — the model processes the entire sentence, so the user does not see each word immediately after typing it.
The project is part of the Digital Brain Project — an initiative to create open neuroscience datasets with a budget of $5 million.
My analysis: Meta's breakthrough in decoding brain activity without invasive methods is a significant step toward commercially available neural interfaces. However, 61% word-level accuracy is still far from practical application. The main challenge is adapting the technology for people with speech and motor impairments, where brain signals may differ significantly from those of healthy volunteers. Keep an eye on this field: in the next 2-3 years, we may see the first prototypes for medical use.