Inside The Black Box: Now Read the Mind of the AI

In this video we open the black box AI and understand, that with euclidean geometry we will not understand the computational complexity of an AI computing on curved manifolds. Luckily we have the mathematical tools and understanding to read the neural mind of the machine. Significant implications for cybersecurity and how to delete knowledge from an AI model. All rights with authors: "Do Sparse Autoencoders Capture Concept Manifolds?" Usha Bhalla⋆ ,a Thomas Fel⋆ Can Rager Sheridan Feucht ,b Tal Haklay ,c Daniel Wurgaft ,d Siddharth Boppana Matthew Kowal Vasudev Shyam Owen Lewis Thomas McGrath Jack Merullo Atticus Geiger† Ekdeep Singh Lubana† ⋆Equal contribution †Equal senior contribution from a Harvard University b Northeastern University c Technion IIT d Stanford University and Goodfire (hint: logo not readable). @harvard @stanford #harvard #stanford #airesearch #aiexplained #newtechnology #scienceexplained