Think about the words swirling around in your head: the bad joke you didn’t say at dinner is the impression your best friend’s new partner made on you. Now imagine that someone can read your mind.

On Monday, scientists at the University of Texas at Austin took a step in that direction. In a study published in the journal Nature Neuroscience, the researchers described an artificial intelligence model that could translate people’s thoughts by analyzing fMRI images. (Functional magnetic resonance imaging)which measure blood flow to different areas of the brain.

To date, researchers relying on implants have developed decoding models for those who have lost the ability to speak and people who cannot write.

According to Shinji Nishimoto, a neuroscientist at Osaka University who was not involved in the study, “Brain activity is a kind of coded signal, and language models provide ways to decipher it.”

But the new language decoder is one of the first to not rely on implants. In research, he was able to transform thought into the form of speech.

The study focused on three participants who came to the lab to listen to podcasts. As they listened, the fMRI scanner recorded blood oxygenation levels in parts of their brains. The researchers then used a large language model to match the patterns of brain activity to the words and phrases the participants heard.

Scientists translated fMRI images into words and phrases.

The result suggests that the AI ​​decoder recorded not only words, but also values. “Linguistic perception is an externally controlled process, while imagination is an active internal process,” the doctor said. Nishimoto.

However, this method of language decoding had its limitations. First, fMRI scanners are bulky and expensive. In addition, the participants in the experiment were able to distract the decoder by thinking about other things.

AI can read our minds, but for now it will have to read them word for word and with our permission.

Source: New York Times.