For years, scientists have attempted to harness the powers of the human brain, but mind-reading has remained firmly in the realms of science fiction and fantasy.
However, a Singaporean study team is creating a basic mind-reading system that decodes brain scans to duplicate visuals that a person is mentally picturing, utilizing artificial intelligence and scanning equipment.
It has the potential to not only transform people with disabilities’ communication skills, allowing them to convey a message with their minds, but it also raises urgent ethical and legal concerns about how the technology could evolve and potentially be misused for corporate interests or surveillance by authoritarian powers.
Zijiao Chen, a Ph.D. student at the National University of Singapore (NUS), is a member of a multinational team that includes people from the United States and China.
She compared their mind-blowing invention to a “mini GPT for the brain,” which, similarly to ChatGPT, leverages a large-scale dataset from a functional magnetic resonance imaging machine (fMRI) “to learn how our brain interprets and thinks about things.”
The method employs brain scans of young adult participants who enter an fMRI for 18 hours and view at images from a dataset of 160,000 images for nine seconds each. The photographs span from everyday structures to sports activities such as baseball and wildlife such as giraffes and swans.
The signals are then processed by “MinD-Vis,” an AI model, to teach it to link specific brain patterns with image properties such as color, shape, texture, and semantics.
In the last stage, it can recover previously undetected visual stimuli from fresh images displayed to participants in the fMRI using brain activity analysis.
“In other words, Mind-Vis is able to read and reconstruct images from our minds,” Ms. Chen explained, adding that the decoded images were consistent. While not 100% accurate, they are recognizable matches that greatly outperformed earlier tests of a similar sort.
“Decoding obtains critical information that aids in understanding how our brain processes and interprets the world around us.” It allows researchers to “visualize and unlock the mystery of the brain, as well as gain a deeper understanding of its complex functions,” she said.
Technology has the potential to improve the lives of the disabled
Within a few years, technology may allow people with impairments to engage with a more advanced machine that can better simulate human cognitive and decision-making processes in real time.
“For example, if somebody is unable to type, they can just picture a sentence… We can decode a sentence he is thinking about. “That’s the long-term goal,” she explained.
Helen Zhou, an associate professor at the NUS Centre for Sleep and Cognition, anticipated that fine-tuning the technology to achieve this goal could take a decade.
She added that for the notion to be commercialized, it would also necessitate a change away from cumbersome equipment like fMRI machines and towards a portable device, such as an EEG (electroencephalogram) headband that utilizes sensors to monitor brain activity.
The goal would be to create a computer capable of real-time decoding that a consumer could take home, connect to WiFi, and readily customize to their specific needs. “That’s the dream,” Ms Zhou added.
She did, however, admit that safeguards were required to prevent such technological improvements from being exploited.
“We really need to be very cautious about governing the usage of this technology and have certain guidelines and firewalls,” she said.
It would be difficult to forcefully read someone’s mind in its current form because the technique requires agreement to function. The individual may thwart the technology simply by training their mind to think about something other than what is being perceived.
However, as science, AI, and technologies become more advanced, the chance of a machine reading someone’s mind without their permission increases.
