Scientists have reportedly created an artificial intelligence-powered bilingual brain implant that turns a stroke survivor's brain activities into sentences, whether in Spanish or English.
Thanks to the years-long work of almost a dozen experts from the University of California, San Francisco's Center for Neural Engineering and Prostheses, the man's brain activity could be translated into sentences in both languages and displayed on a screen.
The individual is identified as Pancho by an article describing their results that was published in Nature Biomedical Engineering on May 20. Following a stroke in the early 2000s, he was left severely crippled at the age of twenty. Though he is unable to speak clearly, Pancho may groan and grunt. As an adult, he acquired English language skills, initially speaking Spanish.
According to the study, this breakthrough represents a significant advancement, as previous efforts to decode speech from brain activity have only succeeded in decoding one language.
In February 2019, Pancho underwent a neural implant under the supervision of Dr. Edward Chang, a neurosurgeon who also serves as co-director of the Center for Neural Engineering and Prostheses. This enabled scientists to begin monitoring Pancho's brain activity.
According to NBC, researchers trained Pancho's implant to decipher words based on the brain activity generated when he tried to speak them, using an AI technique called a neural network. Through AI training, the brain implant, often referred to as a brain-computer interface device in scientific circles, can process data like the human brain.
By 2021, technology had made great strides toward helping Pancho speak again, if only in English. Nonetheless, Pancho's brain implant later became bilingual in Spanish and English because of the decoding system developed based on the study from 2021.
AI Decoding Bilingual Brain Activity
Years after Pancho was paralyzed, the scientists found that his brain retained "cortical activity" in both languages. This finding allowed them to develop a bilingual brain implant instead of training separate language-specific decoding systems.
Scientists set out to demonstrate that in 2022. Once more, they used an artificial neural network to train Pancho's brain implant on the distinct neural activity that his bilingual speech produced.
Their results showed that Pancho could participate in a discussion using the bilingual decoding system that powered his brain implant, alternating between [both] languages when he felt like it.
AI Restoring Voices
Pancho is just another example of how artificial intelligence continues to be used to restore communication for unfortunate patients. Most recently, a patient named Alexis "Lexi" Bogan lost her voice due to an operation involving the removal of a life-threatening tumor near her brain.
August saw the successful removal of a potentially fatal tumor from the back of her brain by surgeons. A month later, after the breathing tube was removed, Bogan struggled to say and swallow words, finding it difficult to even say "hi" to her parents.
Her speech is still affected after months of intense rehabilitation, making it difficult for friends, relatives, and even strangers to understand her attempts at communication. However, Lexi changed in April and began to sound more like herself.
However, it was an artificial intelligence-produced synthetic voice accessible through a mobile app rather than her voice. Using a short audio clip from her adolescence, the AI produced an incredibly lifelike voice.
Sources claimed she can now direct the software to speak anything she wants using her artificially realistic-sounding voice by simply tapping her phone. The clip originated from a school project's cooking demonstration film.
Bogan and many medical experts from the Lifespan hospital network in Rhode Island are hopeful about a prospective application that justifies the possible hazards. Bogan is the sole person with her ailment and one of the first examples of successfully restoring a lost voice utilizing OpenAI's most recent voice engine technology.