An AI “mind-reading” tool can reconstruct what you’re looking at from a brain scan

Researchers at the Weizmann Institute of Science, led by Michal Irani, have developed an AI tool capable of reconstructing visual stimuli from brain scans, a significant advancement in cognitive neuroscience. This tool was presented at the Cognitive Computational Neuroscience conference in New York last month, showcasing its ability to predict what a person is seeing with only one hour of fMRI data, a stark contrast to the typical 40 hours required by previous methods.

The study involved showing participants 9,000 images, with the AI decoder trained on a dataset where 70% of the images were not originally paired with fMRI scans. This innovative approach allows for a more efficient and cost-effective analysis, as fMRI imaging can range from $600 to $1,000 per hour. The high-resolution fMRI voxel used in this study measures just one cubic millimeter, covering approximately 16,000 neurons, which enhances the precision of the reconstruction process.

Tommy Sprague, a neuroscientist at the University of California, Santa Barbara, remarked on the impressive nature of the results, indicating a significant leap in the feasibility of reconstructing thoughts compared to a decade ago. Irani noted that their method outperformed existing techniques by a significant margin, emphasizing the tool's potential applications in therapeutic settings for individuals with neurological conditions, as highlighted by neuroethicist Judy Illes from the University of British Columbia.

However, the research also raises ethical concerns. Marcello Ienca, a neuroscientist and philosopher at the Technical University of Munich, cautioned that while the research is well-intentioned, it could be misappropriated for ethically and societally problematic commercial uses. This duality of potential benefits and risks underscores the importance of careful consideration in the deployment of such advanced AI technologies.