Meta unveils AI models that convert brain activity into text with unmatched accuracy

Meta's AI models decode brain signals into text with high accuracy, using MEG/EEG techniques.

: Meta, in collaboration with international researchers, has developed AI models that translate brain activity into text with unprecedented accuracy. The first project, by Meta's FAIR lab, uses MEG and EEG recordings from 35 volunteers to decode brain signals into typed sentences. The second study maps how thoughts convert into language, showing the brain's dynamic neural code in action. Despite the promise, challenges like MEG's practicality and decoding accuracy remain before clinical application.

Meta, along with international researchers, has made significant strides in AI technology by creating models that transform brain activity into text with remarkable precision. The first study, conducted by Meta's FAIR lab and the Basque Center on Cognition, focuses on decoding brain signals into typed sentences through MEG and EEG recordings from 35 healthy volunteers.

The AI system employs a three-part architecture: an image encoder, a brain encoder, and an image decoder, to align MEG signals for image generation. This method has achieved the ability to decode 80 percent of typed characters, opening new prospects for non-invasive brain-computer interfaces that can aid those unable to communicate.

The second study investigates the neural processes that convert thoughts into words, discovering the brain's use of a dynamic neural code for language production. Although the technology displays potential, challenges such as MEG's size, cost, and environmental requirements persist, necessitating further research for clinical applicability.

Brain decodingNEUROTECHMEG