A few months ago, I started following the amazing Janelle Shane who works with electrical engineering, optics, and neural networks (AI weirdness)…there are some spooky / hilarious GAN images (it can really go either way), absurd character names, recipes (for disaster?), knock-knock jokes, and so much good info to follow on how these networks are trained and refined…the brilliant solutions they come up with and the mistakes they make. Some of the best fails end up here.
Also, I read the news and I’ve been thinking about how deep biases and specifically racist tendencies can go in humans, and how that might show up in AI. If we can refine those algorithms and decision-making biases in the neural networks (which are, after all, loosely based on human brain connections), maybe we can redirect some human malfunction too.