Scientists from Israel have developed an innovative AI system capable of interpreting brain activity scans to recreate images that individuals have viewed. The system employs a decoder model that was trained on open-source data obtained from eight participants, each of whom examined approximately 9,000 images while situated in an MRI scanner. Utilizing the brain scans, the model learned to anticipate the arrangement of colors within an image and its overall content, such as "bananas on a plate."
To create the final image, a separate model, known as the encoder, addresses the inverse challenge by predicting how the brain will respond to a specific image. Both models are trained simultaneously; the encoder produces a predicted scan for a new image, while the decoder works to reconstruct the corresponding original image. This approach enables the system to be trained on images that have never been seen by individuals in an MRI scanner.
Looking ahead, researchers aspire to enhance their capabilities to "read minds" through AI, potentially transforming thoughts and dreams into visual or auditory forms. This technology could provide insights into phenomena like flashbacks associated with PTSD. However, some experts express concerns regarding the ethical implications of such advancements, highlighting the risk of accessing individuals' thoughts without their permission.
Tommy Sprague, a neuroscientist at the University of California, Santa Barbara, cautions that if there exists a method to secretly extract thoughts, the implications could mirror those depicted in science fiction, raising significant concerns. Would you be comfortable with AI interpreting your thoughts?
Informational material. 18+.
