As more people use artificial intelligence (AI), scientists are constantly working on ways to enhance its capabilities. Now, in a new study, researchers have developed a portable, non-invasive system that can decode thoughts and turn them into text.
The technology, developed by researchers from the GrapheneX-UTS Human-centric Artificial Intelligence Centre at the University of Technology Sydney (UTS), aims to improve communication among people who are unable to speak due to illness or injury, including stroke or paralysis. It could also enable communication like never before between humans and machines, such as the operation of a bionic arm or robot.
In the study, participants silently read passages of text while wearing a cap that recorded electrical brain activity through their scalp using an electroencephalogram (EEG). The EEG wave is sorted into different units that capture specific patterns from the human brain, the university’s press statement explained. An AI model called DeWave developed by the researchers is used for this process. DeWave translates EEG signals into words and sentences by learning from large quantities of EEG data.
“It is the first to incorporate discrete encoding techniques in the brain-to-text translation process, introducing an innovative approach to neural decoding. The integration with large language models is also opening new frontiers in neuroscience and AI,” lead author CT Lin said in the statement. The study was conducted on 29 participants and was presented at the NeurIPS conference in New Orleans.
Lin added that this technology is a pioneering effort to translate raw EEG waves directly into language. Previously, the process of translating brain signals to language either required surgery to implant electrodes in the brain, such as Elon Musk’s Neuralink, or scanning in an MRI machine, which is expensive and not something that could work in daily life. Furthermore, the new technology is able to translate with or without eye-tracking unlike previous technology, the researchers explained.
The new model does a better job of matching verbs than nouns. Regarding nouns, the researchers observed a tendency towards synonymous pairs rather than precise translations, such as ‘the man’ instead of ‘the author’.
Previous studies have also attempted to use AI to turn thoughts into text. For instance, in June, a semantic decoder AI was developed by researchers at The University of Texas at Austin that can translate brain activity into continuous text. The study, published in Nature Neuroscience, aimed to capture the summary of what people were thinking instead of word-to-word translation.