Designing human experiences in the age of AI.
Links curated by Peter Polgar.
For a weekly newsletter of hand selected links, signup here.
Posted on 2017 November 6 by polgarp · Tagged in Research and Adversarial attacks
Original link: http://www.labsix.org/physical-objects-that-fool-neural-nets/
This is troubling research. Most AI tech applications can be designed with well-intentioned users in mind, if an error happens, no harm done. But as we replace more and more safety critical applications with AI tech, such attacks may fool systems designed to have better performance than humans.