In a groundbreaking development, researchers at Sydney’s University of Technology (UTS) are working on an artificial intelligence system that translates brainwaves into text—essentially turning thoughts into words.
Led by PhD student Charles (Jinzhao) Zhou and supervised by Chin-Teng Lin and Dr. Leong, the team has built a deep learning model that interprets signals from a 128-electrode EEG (electroencephalogram) cap. When Dr. Leong wore the cap without speaking, the model successfully generated the phrase: “I am jumping happily, it’s just me.”
The AI has been trained on a limited set of words and sentences to streamline recognition and reduce interference from overlapping brain signals. While similar in concept to Elon Musk’s Neuralink, the UTS project is entirely non-invasive, offering a safer alternative to implant-based methods.
“We can’t get very precise because, with non-invasive tech, we can’t tap directly into the brain’s speech-decoding area,” Lin noted.
This technology could revolutionize stroke rehabilitation, autism-related speech therapy, and restore communication for patients with paralysis.
Globally, AI and EEG are being combined for similar breakthroughs. In April, Mass General Brigham researchers unveiled an AI tool capable of predicting cognitive decline years in advance by analyzing brain activity during sleep—achieving 85% accuracy in identifying at-risk individuals.