According to reporters in Tokyo, Japanese scientists recently announced that for the first time, artificial intelligence (AI) can recreate images from human brain activity with an accuracy exceeding 75%.
Scientists have demonstrated that it is possible to recreate all types of images to some extent based solely on thought. (Illustrative image: CCO).
Previously, recreating images from brain activity could only be achieved when subjects were looking at those images, or when the images—such as faces, written words, or simple drawings—were very clear. A team of researchers from the National Institute of Quantum Science and Technology (QST) and other organizations has shown that it is possible to recreate all types of images, including landscapes and complex drawings, to some degree based solely on thought.
Initially, the research team used Magnetic Resonance Imaging (MRI) to record the brain activity of subjects who viewed 1,200 different images. They then employed AI to create a “point chart” with approximately 6.13 million factors such as color, shape, and texture. A neural signal decoding program was created to match brain activity with the point chart, generating new point charts as new brain activity was inputted.
Next, the subjects were shown another set of the 1,200 images they had previously viewed, and their brain activity was measured using MRI 30 minutes to one hour later, while they were asked to imagine the type of image they had seen. Upon inputting the records, the neural signal decoder created point charts. These charts were then fed into another AI synthesis program to recreate images, undergoing a modification process of 500 steps.
This process allows for the identification of the original image from the recreated images with an accuracy rate of up to 75.6%, a significant advancement compared to previous methods that achieved only 50.4% accuracy. The results were published in the online edition of the international scientific journal Neural Networks.
This research could lead to new forms of communication that do not rely on words. Researcher Kei Majima from QST stated: “This is a monumental achievement as humans are seeing into the thoughts of others for the first time. I hope this research will promote a deeper understanding of the human mind.”
