![](https://www.exhibit.tech/wp-content/uploads/2023/03/Self_learning_AI_2880x1620_Lede-scaled-1-1140x694.jpg)
AI has been the talk of the town lately thanks to ChatGPT bringing it to the forefront. AI is now capable of generating human-like conversations which in itself is a trip, but that’s not all as the technology is helping automate various tasks in day-to-day life. Sure, there is a flip side to this AI development as concerns about it replacing many human jobs stand strong. However, its jaw-dropping abilities in the current scenario are worth acknowledging. Just when humanity was catching up to AI being able to have casual conversations with real people, researchers have now revealed that the technology can generate images based on people’s brain activity!
AI images from brain activity
Sounds like science fiction, but researchers have successfully trained an artificial intelligence system to recreate images people have looked at based entirely on brain scans. Researchers Shinji Nishimoto and Yu Takagi from Japan’s Osaka University have recreated Hi-Res images from scans of brain activity. The study held at Osaka University suggests that AI algorithm processed information gathered in different regions of the brain that are responsible for perceiving images. The system interpreted information from the occipital and temporal lobes of the brain via fMRI (Functional Magnetic Resonance Imaging) scans of the brain.
When people look at any image, the temporal lobes register information about its content, whereas the occipital lobes record layouts and perspective. All of this information was recorded using an fMRI that can detect the changing blood flow to the active regions of the brain. The recorded information can then be converted into an imitation of the image according to the researchers. The study was backed by an online data set provided by the University of Minnesota which consisted of brain scans of four people who each viewed over 10,000 pictures.
While this is a big achievement in the world of AI and the technology shows a lot of promise, it still has some limitations. The AI can only recreate images or objects included in its training material which includes data from only four people. Broadening the module will require it to include individual brain scans which is a time-consuming process. As of now, this technology will not be widely accessible anytime soon.