It is now possible to transcribe a person's thoughts by using mind-reading technology, which detects blood flow in the brain in real-time. Researchers said they use artificial intelligence (AI) to analyze fMRI scans to interpret the private thoughts of humans.
The study, titled "Semantic Reconstruction of Continuous Language From Non-Invasive Brain Recordings" published in the journal Nature Neuroscience, conducted an experiment with three participants for the first time, producing a full text of their thoughts instead of only single words or sentences without requiring a brain implant.
Noninvasive AI System Decodes Brain Activity
The team showcased a new noninvasive AI system that can translate brain activity into text. Unlike previous language-decoding methods, The New York Times reports that this system does not rely on implants and can translate imagined speech into actual speech. It can also generate relatively accurate descriptions of what is happening on screen while people are watching silent films.
They focused on three participants who listened to podcasts while undergoing an fMRI scan to record the blood oxygenation levels in their brains. Then, they used a large language model to match patterns in the brain activity to the words and phrases that the participants had heard. The system was able to capture not just words, but also meaning.
The system is a result of large language models like OpenAI's GPT-4 and Google's Bard, which are trained on vast amounts of writing to predict the next word in a sentence or phrase.
These models create maps indicating how words relate to one another, and Dr.Alexander Huth, a neuroscientist at the university who helped lead the research, noticed that particular pieces of these maps could be used to predict how the brain lights up in response to language.
The process was effectively reversed when the researchers used another AI to interpret the individuals' fMRI pictures into words and sentences. They tested the decoder by having the participants listen to new recordings, then seeing how closely the translation matched the actual transcript. The result suggests that the AI decoder was capturing not just words, but also meaning.
Mind-reading AI Raises Concerns of Violation of Mental Privacy
Greta Tuckute, a neuroscientist at the Massachusetts Institute of Technology who was not involved in the research, said that the ability to decode meaning from the brain is a high-level question. She noted that the study showed that it was possible to decode meaning from the brain and that the brain uses common representations across externally driven and active internal processes.
But the technology is also limited to only capturing the main points and not exactly replicating stories, as per MailOnline. Most importantly, the breakthrough raises concerns about mental privacy because it could be the first step in technology listening in on an individual's private thoughts.
Moreover, there were concerns about the technology being used without the individual's knowledge by an authoritative government or employer. However, researchers assured that the system could only read an individual's thoughts after being trained on their specific thought patterns and could not be secretly applied.
Dr. Huth further explained that individuals could control what the technology decoded by simply thinking about something else, which would disrupt the decoding process.
RELATED ARTICLE: Brain Computer Implant Developed for Paralyzed ALS Patients to Control Communication Devices
Check out more news and information on Artificial Intelligence in Science Times.
