Shulou(Shulou.com)11/24 Report--

Beijing, May 2 (Xinhua) scientists have developed a non-invasive artificial intelligence (AI) system that focuses on translating human brain activity into a series of texts, according to a peer-reviewed study published Monday in the journal Nature Neuroscience.

The system, called a semantic decoder, may eventually benefit patients who lose their ability to communicate due to stroke, paralysis or other degenerative diseases. Previously, researchers have developed language decoding methods that allow people who have lost the ability to speak to try to speak and paralyzed people to write when they want to write. But the new semantic decoder is one of the first decoders that do not rely on brain implants. In this study, it can transform a person's imaginary language into actual language, and can produce a relatively accurate description of what is happening on the screen when watching silent movies.

Researchers at the University of Texas at Austin used a "transformation model" (Transformer) in part to develop the system, which is similar to the language model that supports OpenAI chat robot ChatGPT and Google chat robot Bard. Participants in the study trained the decoders by listening to podcasts for hours on functional magnetic resonance imaging (fMRI) scanners. The fMRI scanner is a large machine that measures brain activity.

The scientists recorded MRI data from three participants who listened to narrative stories for 16 hours to train models to map between brain activity and semantic features to capture the meaning of certain phrases and related brain responses.

"it's not just a verbal stimulus," said Alexander Huth, a neuroscientist at the University of Texas who helped lead the study. "We're listening to ideas about what's going on. What could happen is very exciting."

Once the AI system is trained, it can generate a series of texts when participants are listening or imagining telling a new story. The generated text is not an accurate written record, and the researchers designed it to capture general ideas or opinions.

According to a press release issued by the University of Texas at Austin, the text generated by the trained AI system is very close to or exactly matches what the participants are trying to say about half the time. For example, when participants heard the phrase "I don't have a driver's license yet" in the experiment, the idea was translated into "she hasn't even started learning to drive yet."

"for the non-invasive approach, this is a real leap compared to the previous approach, which usually can only generate words or short sentences," Huth said in a press release. "We are asking this model to decode continuous languages with complex ideas for a long time."

Participants were also asked to watch four videos without audio in the scanner, and the AI system was able to accurately describe "certain events", according to the press release.

Limitations, however, Dr. Huth and his colleagues point out that this method of language decoding has limitations. First of all, fMRI scanners are large and expensive. In addition, the training model is a long and tedious process, and individuals must be trained in order to be effective. When researchers tried to use a decoder trained on one person to read another's brain activity, it failed, suggesting that each brain has a unique way of expressing meaning.

As of Monday, the decoder could not be used outside the laboratory environment because it relied on fMRI scanners. But the researchers believe it could eventually be used through a more portable brain imaging system. The main researchers of the study have applied for a Patent Cooperation Treaty (PCT) patent for the technology.

In addition, participants were able to block their inner monologues and get rid of the decoder by thinking about other things. AI may be able to read human thoughts, but at present it can only be read one at a time with human permission.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.