I recently watcher a YouTube video comparing AI safety to Pascal’s mugging, and I quickly felt a whole lot less alone in the world. Robert Miles leans heavily towards AI safety, and I guess I lean into it a little bit (which makes sense, as Robert Miles is an AI research, and I am some asshole), but my refutation comes from the degree with which I think it is needed right now (vs whether it is needed at all). I do think it is an overpopulation on Mars type problem, and this has led to some really fucking irritating conversations.
Book’s like Janelle Shane’s You Look Like a Thing and I Love You are a good primer for my opinion.
As far as I can tell, Janelle Shane wanted to put a cute and cuddly face on Artificial Intelligence. You get that from the adorable drawings that help explain some of the concepts of this book. If those two sentences feel like I am being dismissive of the author, I am not. I am firmly in the camp that of using the teaching tools that work, and I think Shane really gets what works. This is matched by a clear and legible prose which really works, and really works well.
People love to throw around the Arthur C Clarke quote about a sufficiently advanced technology being indistinguishable from magic and most people misunderstand that quote as to be really cool science fiction fodder. I have a different relationship with that quote, as I spend at least an hour of my week explaining to my father that just because his computer is slow doesn’t mean the CIA is spying on him. Technology is a series of windmills that breeds Don Quixote of us all, and that’s how I see AI. In the modern world, AI is largely a marketing buzzword (like ‘organic’) that has become semiotically decoupled from its signified, and thus become three different signs; for computer scientists it is one meaning a process by which certain computer programs function (with comical results, as the the title of this book suggests): for marketers, a buzzword meaning high-tech: for the population at large, skynet. As far as I am concerned, the only group who’s opinion matters is the first, and Janelle Shane is firmly in that catagory.
If you want to have competent opinions on what is actually going on in the world around you, I recommend this over Nick Bostrom.
I don’t want to make this blog post too much about my opinions, SkyNet, Pascal’s Mugging, or all the things that aren’t actually present in Shane’s book. But I will close by saying that when I tell people that I am not worried about the eminent coming of SkyNet, books like this are what have informed my opinion on what AI actually is.
Failures in Polymathy 