You know I'm actually an AI enthusiast (computer science graduate with specialization in artificial intelligence) so I'm quite on the fence about this. First of all, let me just acknowledge that the video was a good explainer, especially for the layman. The concepts were introduced effectively, and the sequence of the topics was arranged neatly. I do agree (and often preach) that we are too far off from a robot apocalypse that sci-fi stories narrate. Though, I do have qualms about how Clif argued it.

During my thesis, I was actually part of a laboratory dedicated to giving AI the ability to recognize and, hopefully, emulate human emotions. The main goal of it was to assist geriatric patients and people who suffer from mental conditions. With that being said, I've had more than a year to digest literature surrounding the topic. While I'm not nearly as smart or experienced as Clif and other experts, I do have a basic sense of the field. Consider me a minnow in the sea of AI, so take my opinion as such.

It's true that machines, in its most primal state operate on line by line programming, but we can't just talk about it in terms of low-level programming. Not anymore. While their core operates on 1s and 0s, computers have evolved so much these past few years. Deep learning, neural networks and fuzzy logic pave the way for computers to "think" the way humans think. Given the advancements in technology, one might argue that in a few years the capability for them to "think" better is not far off. I use quotations with think, because they don't really have free will or even a "soul", but there have been developments in trying to instill them with such, even before I even started learning how to program.

Machines do have the ability to reprogram themselves, even though with a limited capacity. Will they be able to reprogram their core directive? It's up to the safeguards that their developers put in place. Machines that are able to learn unassisted already exist, but their reach is limited because that's how they're programmed. They won't be taking over the world any time soon, but they do learn and think, not like humans though, but they still have the ability to do so. Because, that's how they were made.

Sorry for the long comment, man. I have too much more to say, but I don't want to hijack your comment section. Great topic for discussion though.