I believe there is a great misunderstanding with the current position of artificial intelligence and its possible applications. There are two stereotyped groups: those who understand the technology deeply and those who understand the practical realities of the world. Some people understand both, but often they still sensationalize as they are excited by the applications of the technology.
Without an understanding of computer science, it is difficult to make a judgment if AI is paradigm shifting or incremental in nature. I believe that current AI technologies, primarily learning algorithms, fall somewhere in-between. They will disrupt the economy and result in productivity gains, which is good news. This disruption will be asymmetrically distributed in industries where the work and problems are conducive to this technology’s application. Therefore, like many times in history, the job disruption will be concentrated in a relatively small proportion of industries and employees. For example, self-driving cars are an algorithmic problem that is a perfect match for machine learning. This does not mean that machine learning is universally applicable.
The following video from Vox does a good job bridging the gap and teaching what are reasonable applications given the current state of AI.