An artificial intelligence can decode （解码） words and sentences from brain activity with1 surprising accuracy. Using only a few seconds of brain activity data, the AI guesses what a person has heard. It lists the correct answer in its top 10 possibilities , researchers found in a primary study.
Developed at the parent company of Facebook, Meta, the AI could eventually be used to help thousands of people around the world who are unable to communicate through speech, typing or gestures.
Most existing technologies to help such patients communicate require risky brain surgeries to put in electrodes （电极）. This new approach "could provide a possible path to help patients with communication problems, avoiding the use of surgeries, " says neuroscientist Jean-Rémi King, a Meta AI researcher.
King and his colleagues trained a computational tool, also known as a language model, to detect words and sentences on 56, 000 hours of speech recordings from 53 languages. The team applied an AI with this language model to databases from four institutions that included brain activity from 169 volunteers. In these databases, participants listened to various stories and sentences, while the people's brains were scanned by magnetoencephalography （MEG）（脑磁图）.
Then with the help of a computational method that helps account for physical differences among actual brains, the team tried to decode what participants had heard using just three seconds of brain activity data from each person. The team instructed the AI to match up the speech sounds from the story recordings with patterns of brain activity that the AI computed as corresponding to what people were hearing. It then made predictions about what the person might have been hearing during that short time, given more than 1, 000 possibilities. Using MEG, the correct answer was in the AI's top 10 guesses, the researchers found.
"The new study is decoding of speech recognition, not production, " King agrees, "Though speech production is the final goal, for now, we're quite a long way away. "
