In 2017, Facebook (now Meta) had a bold idea: a brain-reading hat that would let you type just by thinking. Years later, the company actually built something similar, but the catch is that it’s nowhere near ready for everyday use.
Meta’s brain-typing system uses artificial intelligence (AI) and neuroscience to analyse brain activity and predict which keys a person is pressing, just from their thoughts. But there’s a catch: the system requires a massive, expensive machine and can only work in a highly controlled lab setting.
How Meta’s brain-typing AI works
According to a recent blog by MIT Technology Review, the tech relies on a special brain scanner called a magnetoencephalography (MEG) machine, which detects tiny magnetic signals created by brain activity. The scanner is so large and sensitive that it needs to be kept in a special room to block interference from Earth’s magnetic field.
Meta’s researchers trained an AI model, called Brain2Qwerty, to analyse these brain signals. The AI learned to match patterns in the data to specific letters as volunteers typed on a keyboard. Over time, the system became accurate enough to correctly predict which letter a person was thinking about 80% of the time.
The experiment involved 35 volunteers at a research centre in Spain. Each person spent about 20 hours inside the scanner, typing sentences while the AI studied their brain activity.
The Challenges: Why this won’t be a product anytime soon
Despite its progress, Meta’s brain-typing system is far from becoming a consumer product. There are several reasons for this: • Huge and expensive equipment: The MEG scanner weighs half a ton and costs around $2 million, making it impractical for everyday use. • Requires a completely still head: If a person moves even slightly, the brain signals become unreadable. • Not accurate enough yet: While the system is the most advanced of its kind, it still gets about 32% of letters wrong.
Meta’s research team, led by Jean-Rémi King, is not trying to create a product but instead hopes to learn more about how human intelligence works. Understanding the way the brain structures language could help improve AI systems, such as chatbots and language models.
Although Meta’s brain-typing AI isn’t ready for the real world, brain-computer interfaces (BCIs) are advancing quickly. Some of the most effective systems today use electrodes implanted in the brain, which allow paralyzed patients to control computers or even “speak” through a synthesised voice.
Companies like Neuralink, founded by Elon Musk, are developing brain implants that could restore movement and communication for disabled people. While Meta’s research focuses on non-invasive methods (without surgery), it still has a long way to go before brain-controlled typing becomes practical.
For now, Meta’s brain-typing AI remains an exciting scientific achievement—but one that’s likely to stay in the lab.