U.S. scientists develop an AI-powered implant to decipher thoughts
Stanford University scientists have developed an AI-based brain implant that deciphers thoughts, enabling people with speech impairments to vocalize their ideas.
For the experiment, participants had electrode arrays implanted in the motor region of their brain to monitor neural activity. AI models, trained to decode these neural signals, linked them to individual speech units and then to full sentences. The researchers emphasize that the use of artificial intelligence was key to achieving a very high accuracy in deciphering the brain's signals.
A History of Innovation Stanford scientists have been working on these neural interfaces since 2017, when they conducted their first studies using brain implants. One experiment became a global sensation: a paralyzed man had two implants, each with a hundred electrodes, embedded in his motor cortex. The implants read signals from 200 neurons at once. To "write" a letter on a screen, the man would imagine making the hand movement to do so. The implants then transmitted the signals to an algorithm, which translated them into letters that appeared on the screen.
In the following years, researchers refined the technology. In 2024, they developed a brain implant that allowed paralyzed people to convert their thoughts into words on a display with an accuracy of about 90%. A person was able to communicate at a rate of 62 words per minute, which is the speed of a normal conversation.
Now, the use of AI has not only improved the system's accuracy but also made its performance exceptionally stable.
ORIENT
