1. AI Is As Smart As We Can Code It Via Twitter: @cbrisket And you can only say so much based on the programming languages we have at our disposal. Paul Graham has written quite eloquently on this subject, and his point seems to me to be that we don't yet have a software means of harnessing the power of our hardware. 2. Popularity Is Not Progress Via google.com Sure, deep learning is popular in commercial and academic circles, but it's not as though we have robots that can drink wine and discuss Foucault with us. 3. The Qualia Question Via media.salon.com Please read this book. V. S. Ramachandran, the author of Phantoms of the Brain, details here many of the unique facets of the brain we can't yet explain via science. Conscious sensations, or qualia, is one of the parts of being human that defies explanation. If we can't yet explain ourselves, I find the idea of building machines smarter than us a bit putting the cart before the horse, no? 4. About Those Unanswered Questions in Computer Science... Via yaddlezap.com There's a bunch of them that are still outstanding. Understanding any of them at a more comprehensive level could change our thoughts on AI. Similarly, what about all the unanswered questions in neuroscience? Again, it seems premature to be forming an opinion on whether or not AI is dangerous when so many of these questions still exist. 5. Can't Robots Be Friends? Via venturebeat.com Via Business Insider, Heather Knight of Carnegie Mellon University has done extensive research on human-robot social interactions. Knight notes that soldiers that work with bomb disposal robots develop emotional connections with them, and are wary of throwing even broken ones to the scrap heap. Although (at present) that emotional connection is not mutual, software engineers in love are ones more likely to code love back.