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

Beijing time on September 14 morning news, according to reports, at present, artificial intelligence technology makes non-invasive brain decoding technology further! Although the technology doesn't yet enable people who can't talk to each other to talk and laugh like normal people, it does allow scientists to decode their speech accurately.

AI technology can decode words and sentences from participants 'brain activity with incredible accuracy, up to 73 percent accuracy, which means AI is a step closer to "non-invasive decoding" brain activity data for brain language and consciousness!

The AI technology can decode words and sentences from participants 'brain activity with incredible accuracy, but it is still not 100% accurate. It takes only a few seconds of brain activity data to guess what people hear. In a preliminary study, they found that participants found the correct answer from the selective test on average 73%.

Giovanni Di Liberto, a computer scientist at Trinity College Dublin, Ireland, who was not involved in the study, said AI had outperformed what many thought was possible.

Meta, Facebook's parent company, has developed a new artificial intelligence technology that could eventually be applied to tens of thousands of people around the world who cannot communicate via voice, typing or gestures, including patients with minimal consciousness, locked-in syndrome or "vegetative state," now commonly referred to as unresponsive waking syndrome, media reported on Aug.25.

It is reported that most of the current technology to help people with language disorders has a certain degree of physical invasiveness, requiring high-risk brain surgery to implant electrodes, Meta AI researcher, neuroscientist Jean-Remi King (Jean-Remi King) said, the latest development of artificial intelligence technology is expected to provide a feasible solution to help patients with communication disorders, rather than using invasive methods.

Jenny and colleagues developed a computational tool that detects words and sentences in 56000 hours of voice recordings in 53 languages, also known as language models, that learns how to recognize specific features of language at both subtle levels (e.g., letters or syllables) and broader levels (e.g., words or sentences).

The research team applied the AI system with language models to four institutional databases containing brain activity from 169 volunteers, where participants listened to different stories and sentences, such as: Ernest Hemingway's The Old Man and the Sea and Lewis Carroll's Alice in Wonderland, during which participants 'brains were scanned using magnetoencephalographs (MEGs) or electroencephalographs, which measure the magnetic or electrical components of brain signals.

Then, with the help of a computational method that helps explain the physical differences between actual brains, the team tried to decode what they heard using three seconds of brain activity data from each participant, instructing the AI system to match the speech in the story recording with the brain activity patterns that the AI calculated corresponded to what people heard, and then predicting what participants were likely to hear in a short time based on more than 1000 possibilities.

The researchers found that when the MEG was used, participants were 73 percent accurate in choosing the top 10 possible answers, but when the EEG was used, that accuracy dropped to less than 30 percent, so the MEG performed very well. "But we're not optimistic about the actual application of the system," Liberto said."What can it do? MEG is a bulky and expensive machine, and applying the technology to clinics requires technological innovation to make the device cheaper and easier to use.

Jonathan Brennan, a linguist at the University of Michigan in Ann Arbor, said: "In this latest study, it is important to understand the true meaning of 'decoding,' a term often used to describe the process of deciphering information directly from the source of information, specifically from brain activity, and artificial intelligence technology can achieve this because the system can provide a limited range of possible answers with greatly improved accuracy." For language, if we want to extend this AI system to practical applications, this is very difficult to achieve, because language applications are infinite. "

More importantly, AI can decode information about passive listening participants, which is not directly related to nonverbal patients. In order for it to be a meaningful communication tool, scientists need to learn how to decipher the information that patients want to express from brain activity, such as hunger, discomfort expression or simple "yes" or "no" expression.

In fact, this artificial intelligence technology is decoding speech perception, not speech generation, although speech generation is the ultimate goal of scientists, but for now, there is an urgent need to further improve the relevant science and technology.

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.