Call it a Chat-scan.<\/p>\n
Just in case artificial intelligence didn\u2019t spark enough privacy concerns: Israeli scientists have devised a mind-reading program named Brain-IT that can tell what people are thinking just by analyzing their brain scans \u2014 and then recreate that image with shocking accuracy.<\/p>\n
The technological telepath \u201cproduces reconstructions that more closely resemble the viewed images both structurally and semantically,\u201d the researchers wrote in a paper<\/a> submitted to the International Conference on Learning Representations.<\/p>\n A new mind-reading program called Brain-IT can recreate shockingly-accurate visual depictions of people\u2019s thoughts based on their brain scans. Login \u2013 stock.adobe.com<\/p>\n In this image, the left-hand image of each pair is what a user actually saw, and its right-hand counterpart is what a new AI model reconstructed from their brain scan. Weizmann Institute of Science<\/p>\n These precise reproductions entail everything from stop signs to an olive and tomato pizza with the exact number of slices people had envisioned, like the digital equivalent of a counterfeit bill.<\/p>\n They demonstrated these in a series of composites that juxtaposed the original image (left) with the one the AI recreated based on their brain scans (right).<\/p>\n Coincidentally, AI mind readers are nothing new<\/a>. In 2017, researchers at Purdue University in Indiana devised a model that could predict what image a subject was thinking of, but could only produce grainy recreations, Popular Mechanics reported.<\/a> Meanwhile, in 2022, Japanese scientists developed an AI model dubbed Stable Diffusion in 2022 that could more precisely approximate what a subject was but relied on text to image tech.<\/p>\n By contrast, Brain-IT recalls the images to a T \u2014 and in just one hour, rather than multiple hours like prior iterations, Weizmann Institute of Science\u2019s Michal Irani, who spearheaded the tech\u2019s development,\u00a0said in a statement<\/a>.<\/p>\n \u201cWe realized that by translating back and forth \u2013 from a random image that had never been viewed in an fMRI machine, to a predicted brain scan, and then back to the image we started with \u2013 the models would effectively build themselves a massive dataset,\u201d said Irani. Weizmann Institute of Science<\/p>\n Researchers built the system by feeding it over 70,000 images that were collected while eight subjects looked at various pictures.<\/p>\n As this was a relatively small sample size by AI brain scanner<\/a> standards \u2014 generating a larger one would require cramming vast numbers of participants in MRI machines for long periods \u2014 researchers were forced to rely on an \u201cencoder.\u201d<\/p>\n Using this shortcut, the tech could both decode the images from brain scans and vice versa to build a large sample.<\/p>\n \u201cWe realized that by translating back and forth \u2013 from a random image that had never been viewed in an fMRI machine, to a predicted brain scan, and then back to the image we started with \u2013 the models would effectively build themselves a massive dataset,\u201d said Irani. \u201cIn this way,\u00a0during training, the models would learn to generate scans that encode images effectively, even though those fMRI scans had never actually been performed.\u201d<\/p>\n During this mind-blowing experiment, the tech identified \u201c128 functional regions\u201d \u2014 some known and some new \u2014 that perform similar functions in all people. <\/p>\n For instance, some of these image synthesizers responded to images of food, while others were activated by sports pics.<\/p>\n Meanwhile, different areas corresponded to indoor and outdoor thoughts.<\/p>\n The researchers believe that Brain-IT could expedite brain scans and help paralyzed people communicate. Weizmann Institute of Science<\/p>\n Irani said they could use the tech to help people suffering from paralysis communicate, as well as study the brain more efficiently and cost-effectively.<\/p>\n \u201cNone of us can afford 40 hours of imaging for a new subject,\u201d said Tommy Sprague, a neuroscientist at the University of California, Santa Barbara, in a statement. <\/a>\u201cIt\u2019s something like $600 to $1,000 an hour.\u201d<\/p>\n In the future, the tool could be used to probe even further into our psyche, the team hopes.<\/p>\n \u201cIf we overcome all these obstacles, it\u2019s possible that in the future we may even be able to read dreams,\u201d said Irani.<\/p>\n