Meta makes progress towards AI system that decodes images from brain activity

The new system combines a non-invasive brain scanning method called magnetoencephalography (MEG) with an expert system system.A common MEG imaging device. Image public domain, source: NIMH/WikipediaThis work leverages the companys previous work translating letters, words, and audio spectrograms from intracranial recordings. According to a Meta blog site post, “This AI system can be released in real time to rebuild, from brain activity, the images viewed and processed by the brain at each immediate.” A post from the AI at Meta account on X, previously Twitter, showcased the real-time abilities of the design through a presentation illustrating what a person was looking at and how the AI decoded their MEG-generated brain scans.Today were sharing new research that brings us one action closer to real-time decoding of image perception from brain activity.Using MEG, this AI system can decipher the unfolding of graphes in the brain with an unmatched temporal resolution.More information ⬇– AI at Meta (@AIatMeta) October 18, 2023

Its worth noting that, regardless of the development revealed, this experimental AI system requires pre-training on a persons brainwaves. In essence, rather than training an AI system to read minds, the designers train the system to translate specific brain waves as particular images. The group has particularly noted that this research is part of the companys ongoing effort to unwind the secrets of the brain.