I didn't think this would come so soon it's just still 2022. https://www.nature.com/articles/s41598-021-03938-w
TLDR: Based on my skim read but mostly payed attention mostly not skim-read: They trained a GAN on the celeb dataset and used it to generate face images and showed them to some humans (~2000 images), test set first then train set, then used this MRI headset data to train another GAN AI on learning the relations of brain waves <> latent space of already trained AI. With this, they were able to generate those pictures of them thinking (or was it seeing?) of the person, and it does match the original very closely, and besides it is what they thought of i.e. so they basically made art. Yes, they triggered a similar latent space by thinking of the face, but if they think of a close other latent then it would be a bit different face under your control! Now scale this up 1000 times and we have maybe DALL-E 2 for brains and crazy mind uploading and gods, sharing their visions wirelessly to another person through their google glasses and then back out to the sender and repeat. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T17d301092e0442c4-M8620d8e8256d49d689712ed6 Delivery options: https://agi.topicbox.com/groups/agi/subscription
