Researchers have successfully enhanced brain-reading implants with artificial intelligence (AI), enabling people with paralysis to communicate with unprecedented accuracy and speed. Two separate studies published in Nature describe the development of brain-computer interfaces (BCIs) capable of converting neural signals into text or words spoken by a synthetic voice. These BCIs can decode speech at rates of 62 and 78 words per minute, respectively, surpassing all previous attempts.
– Francis Willett: “We can now envision a future in which we can give a person with paralysis back the ability to speak freely, allowing them to freely say whatever they want to say with enough accuracy to be reliably understood.”
The potential to restore normal spoken language
The implications of this discovery are enormous. Neurologist Francis Willett, a co-author of one of the papers, is optimistic about the future, stating, “We can now envision a future in which we can give a person with paralysis back the ability to speak freely whatever they want to say with high enough accuracy to be understood.” This technology has the potential to revolutionize the lives of people with paralysis, allowing them to communicate freely and effectively.
Impressive results and future prospects
The scientists who conducted this research have made significant progress in developing BCIs that can accurately interpret neural activity and translate it into text or speech. Willett’s team implanted arrays of small silicon electrodes into the brain of a patient with motor neuron disease, training deep learning algorithms to recognize the unique signals associated with speech. The AI was able to decode words from phonemes, achieving an error rate of 9.1% for a small vocabulary of 50 words. Although the error rate increased to 23.8% when the vocabulary was expanded to 125,000 words, the AI still correctly decoded about three out of every four words.
– Pat Bennett (patient): “For those who don’t speak, it means they can stay connected to the larger world, perhaps continue to work, maintain friendships and family relationships.”
In another study, neurosurgeon Edward Chang and his team used electrocorticography (ECoG) to record the brain activity of a patient who had lost the ability to speak after a brainstem stroke. They trained artificial intelligence algorithms to recognize patterns in the patient’s brain activity associated with attempts to speak. This approach yielded impressive results: the BCI produced 78 words per minute, and the average error rate in pronouncing words was 25.5%.
The future of brain-reading implants
Experts believe that brain-reading implants could become a product in the near future. Christian Herff, an expert in computational neuroscience, claims that the potential for commercialization is within reach. If further developed and refined, these BCIs could transform the lives of people with paralysis, allowing them to stay connected to the world, maintain relationships and even continue working.
– Christian Herff: “These devices could become products in the very near future.”
