This Japanese AI can recreate what you see by analyzing a brain scan


Researchers at Osaka University have developed an artificial intelligence system for reconstructing images using brain scans.

Neuroscientist Yu Takagi and his research partner Shinji Nishimoto used a model they created along with the 2022 German AI algorithm Stable Diffusion to generate the images by transforming people’s brain activity in an MRI machine.

Stable Diffusion is typically used to transform words and phrases into visual representations. The technology was trained to scan existing images and their captions, eventually learning to make connections between specific images and words.

Takagi and his team integrated their own training with two different AI models into this technology: one capable of combining images with functional magnetic resonance imaging (fMRI) data, and the other capable of combining fMRI data with textual descriptions to link the images.

More from NextShark: Asian-American Founders Launch FAM App to Take on Clubhouse

“I still remember when I saw the first pictures. I went into the bathroom and looked at myself in the mirror and saw my face and I was like, ‘Okay, that’s normal. Maybe I won’t go crazy,'” Takagi told Al Jazeera.

The technology, which received an accuracy rating of about 80%, uses the first AI model to create a vague and unclear image of what participants saw in the MRI machine, and then uses the second model to create the ones previously used Recognizing images and clarifying recorded associations of brain patterns.

“We really didn’t expect such a result,” said Takagi.

More from NextShark: Chinese accused of killing Nigerian ex-girlfriend because she rejected him

The 34-year-old researcher, who is also an assistant professor at the university, stressed that his discovery shouldn’t be taken as mind-reading, as his method can only produce images that a person has seen before.

READ :  Artificial intelligence discovers the 'toledano steel' of the future

“Unfortunately, there are many misunderstandings in our research,” Takagi said. “We cannot decode imaginations or dreams; We think this is overly optimistic. But of course there is potential for the future.”

The story goes on

More from NextShark: Maryland’s First Lady Yumi Hogan secures 500,000 COVID-19 test kits from South Korea

While the breakthrough is noteworthy, it has also sparked concerns and debates about the potential risks it may pose to society, particularly to individual privacy.

Takagi has acknowledged these concerns as valid and recognized that some with malicious intent could attempt to abuse the technology.

“Privacy issues are the most important thing for us,” Takagi said. “If a government or institution can read people’s minds, that’s a very sensitive issue. High level discussions need to be held to ensure this cannot happen.”

More from NextShark: Power Rangers Reunion Special Trailer reveals the fate of the original Yellow Ranger Trini

Takagi and Nishimoto published their findings in December and plan to present their work at the Conference on Computer Vision and Pattern Recognition in June.