Meta's new AI can turn your thoughts into text without a brain implant

Meta has unveiled **Brain2Qwerty v2**, an advanced artificial intelligence system designed to convert brain activity directly into text without requiring invasive brain implants. While the technology remains in the research phase, it represents a significant step toward enabling thought-based communication for individuals affected by stroke, traumatic brain injuries, or neurological disorders that impair speech.

Unlike conventional brain-computer interfaces that depend on surgically implanted electrodes, Brain2Qwerty v2 relies on **magnetoencephalography (MEG)**, a non-invasive imaging technique that captures the brain's magnetic signals using external sensors. This allows the system to interpret neural activity without any surgical intervention. According to Meta, the latest version is its most sophisticated brain-to-text model, capable of reconstructing complete sentences from brain signals with near real-time performance.

To develop the system, Meta's researchers worked with nine participants, each spending approximately ten hours inside an MEG scanner while typing nearly 22,000 sentences. The extensive dataset enabled the AI model to identify relationships between neural activity and written language, improving its ability to translate thoughts into text.

A major advancement in Brain2Qwerty v2 lies in its AI architecture. Instead of relying on manually selected neural features, the system uses an end-to-end deep learning approach that learns directly from raw brain signals. Meta also incorporated large language models to better interpret context, enabling the system to generate more accurate predictions even when brain signals are incomplete or affected by noise. AI agents were further utilized to evaluate multiple optimization strategies before selecting the most effective training pipeline.

The upgraded system demonstrated a substantial improvement in decoding performance. Meta reported an average word accuracy of **61%**, a significant increase over earlier non-invasive brain-computer interface systems, which typically achieved around **8%** accuracy. The highest-performing participant reached **78%** word accuracy, and more than half of the decoded sentences contained no more than one incorrect word. Researchers also observed that increasing the amount of training data further enhanced the model's accuracy, indicating strong potential for future improvements.

Meta emphasizes that Brain2Qwerty v2 is not intended to replace traditional input devices such as keyboards or smartphones. Instead, its primary objective is to provide a non-invasive communication solution for individuals who have lost the ability to speak due to neurological conditions. Although implanted brain-computer interfaces currently deliver higher accuracy, they require complex surgical procedures, limiting their accessibility. Meta believes its non-invasive approach could make brain-to-text communication available to a much wider population.

To support continued innovation in this field, Meta is releasing the complete training code for both Brain2Qwerty v1 and Brain2Qwerty v2. In addition, its research collaborator, the Basque Center on Cognition, Brain, and Language (BCBL), will make the original Brain2Qwerty v1 dataset publicly available. Through these open research initiatives, Meta aims to accelerate the development of brain models, advance neurological research, and help restore communication for people living with speech impairments.

Voice Of Osiz

At Osiz, we believe Meta's Brain2Qwerty v2 marks a significant milestone in the evolution of AI-powered healthcare and brain-computer interfaces. By combining advanced deep learning with non-invasive neural decoding, this innovation demonstrates how AI can unlock new possibilities in assistive communication. As AI models become more capable of interpreting complex brain signals, they will pave the way for accessible, patient-centric healthcare solutions. We see immense potential for AI to accelerate neurological research, improve rehabilitation outcomes, and transform the quality of life for individuals with communication disorders. Open research initiatives like this will further fuel innovation across the healthcare ecosystem. At Osiz, we empower organizations with cutting-edge AI development solutions that drive meaningful advancements in next-generation healthcare technologies.

Source: Firstpost.com