How AI May Soon Be Able to Decode Our Thoughts


Alyssa Rask, Staff Writer

Scientists in Texas may have found a way to use brain scans and AI to get a hint into our thoughts—making mind-reading a real possibility in the future.

On Monday, researchers from the University of Texas published a study in the Journal of Nature Neuroscience, explaining an AI that could translate the thoughts of humans by analyzing MRI scans.

Alexander Huth, one of the top leaders of this study, stated: “For a noninvasive method, this is a real leap forward compared to what’s been done before, which is typically single words or short sentences.”

The system is called a semantic decoder, a non-invasive method that includes measuring brain activity by using the FMRI scanner. The scanner works by imaging blood flow across different parts of the brain.

The University of Texas study entailed 3 people listening to podcasts from headphones for up to 16 hours in the MRI scanner. While this occurs, the semantic decoders are trained through this imaging, which allow them to predict the activity of the brain when hearing certain words or phrases.

Following the training of the decoder, if the patient allows for their thoughts to be decoded, they begin to listen to a new podcast and the machine creates corresponding text from the brain activity. Researchers found that the decoders were able to identify what the study participants were hearing from brain activity.

While this new area of science is exciting, many people have begun to question the ethics involved with the system and the possible future of using it in interrogations. Others believe that the decoder can interfere with mental privacy.

Jerry Tang, a Ph.D. student at the university who worked on the project, stated that while further working on the decoder, they believe it is important to “enact policies that protect each person’s mental privacy.”