A team of researchers has developed an AI system capable of recreating images from brain activity, showcasing the ever-expanding capabilities of artificial intelligence.
While generating human-like responses or hyper-realistic images has become routine, this new AI system marks a significant advancement. Researchers at Radboud University in the Netherlands have engineered an AI system that can accurately reconstruct what a person is seeing by analyzing their brain activity.
These reconstructions improve as the AI learns which brain regions to focus on. According to Umut Güçlü from Radboud University, these are the most accurate reconstructions to date, as reported by NewScientist.
The researchers utilized fMRI scans, which display activity in specific brain areas, to record brain activity while subjects viewed images. They also employed direct electrode recordings from a macaque monkey. Over time, the AI system learned to concentrate on the most relevant brain regions, significantly enhancing the accuracy of the image reconstructions.
Another study involved using implanted electrode arrays to directly record the brain activity of a macaque monkey observing AI-generated images.
Recently, the team reanalyzed data from these previous studies using an improved AI system that identifies which brain regions to focus on. Despite some limitations, such as relying on images already in the dataset, this research is groundbreaking.
It holds potential for significant applications, including aiding stroke victims in communication and interpreting dreams, marking a substantial leap forward for AI technology.
