- Goal is testing a machine that decodes brain signals in words written on a computer.
- The cerebral typing system is up to 80% precise but nowhere practical.
- The machine is half a ton, costs $ 2 million, needs a armored room and even mild head movements interrupt the signal.
Goal is showing a machine capable of turning your thoughts into written words on a screen, but do not wait to write your telepatically Instagram subtitles in the short term. The device weighs approximately a half ton, costs $ 2 million and is as portable as a refrigerator. Therefore, unless you plan to load a Magnetoencence (MEG) laboratory level, you will not send mental text messages in the short term. And that is even before considering how you can’t even move your head slightly when you use it.
Even so, what Meta has done is impressive. Its AI and Neuroscience teams have trained a system that can analyze brain activity and determine which keys someone is pressing, based solely on thought. There are no implanted electrodes, there are no science fiction headbands, only a deep neuronal network decipher the brain waves from the outside. The investigation, detailed in a couple of recently launched documents, reveals that the system is up to 80% precise to identify letters of brain activity, which allows it to rebuild complete sentences of the thoughts of a typing.
While writing phrases, a volunteer is inside a meg scanner, which looks a bit like a giant hair dryer. The scanner collects magnetic signs from neurons shooting in the brain, and an AI model, called Brain2qwerty, gets to work learning what signs correspond to which keys. After enough training, the letters that a person is writing can predict. The results were not perfect, but they could reach precision levels of up to 80%.
Cerebral typification
Telepathic typing has some real limits for now. The scanner must be in a especially armored room to block the earth’s magnetic field, which is a billion times stronger than what is in your head. In addition, the slightest head inclination stir the signal. But there is more than just another goal product. Research could really promote brain science and, finally, medical care for brain injuries and diseases.
“To explore how the brain transforms thoughts into intricate sequences of motor actions, we use AI to help interpret Meg’s signals while participants wrote sentences. When taking 1,000 snapshots of the brain every second, we can identify the precise moment in which They turn thoughts into thoughts. “Our study shows that the brain generates a sequence of representations that begin from the most abstract level of representations, the meaning of a sentence, and progressively transforms them into a myriad of actions, such as the real movement of the finger on the keyboard.”
Despite its limitations, the non -invasive aspect of finishing research makes a much less scary approach that covers a computer chip directly in its brain, since companies like Neuralink are testing. Most people would not enroll in elective brain surgery. Although a product is not the declared objective of research, historical points show that giant and laboratory machines do not have to stay that way. A small smartphone does what a computer of the size of a building could not in the 1950s. Maybe today’s brain scanner is tomorrow’s laptop.
