Also the abstract was a poor definition of the interesting points of the paper. This was one of the earlier papers that raised the risk and dangers of such a technology. People tend to say "AI is dangerous" but this actually defines where and how the harm comes in.