Meta has created a device that enables users to produce text by simply thinking what they want to say. Researchers used a state-of-the-art brain scanner and deep learning AI model to interpret neural signals, achieving an accuracy of 80% in detecting keyboard keys.
The device uses magnetoencephalography (MEG) technology, which detects magnetic signals in the brain without requiring invasive implantation. However, this method comes with significant limitations, including a large and expensive scanner weighing half a ton and costing $2 million.
Additionally, the system requires users to be stationary, as any head movement can disrupt the signal. These caveats make commercialization challenging. Nevertheless, Meta views the achievement as an impressive milestone in understanding human brain architecture and developing AI models.
The Brain2Qwerty system learns from observing keyboard inputs over several thousand characters, achieving a 32% error rate. While not perfect, this is the most accurate brain-computer interface device using a full keyboard that reads signals from outside the skull.
Meta’s research confirms the theory of hierarchical language formation in the mind, which could be beneficial for AI development. As King notes, “Language has become a foundation of AI,” and understanding computational principles behind brain acquisition is essential.
While this technology is unlikely to lead to practical applications directly, the researchers are excited about their findings, which could pave the way for future breakthroughs in AI research.
Source: https://futurism.com/neoscope/meta-device-type-with-brain