The first thing I feel I need to point out is that my horse in this race is not whether or not jobs will go away, but whether or not the original commenter to whom I was replying advanced the discussion at all with their remark. I think that the remark–that people have always been saying that automation will take over but it never has–is facile in that the thing that "people have been saying" is not the same over time. In the late 18th century, it could only really have been a comment about capital concentration and the death of the trades. In the 1950s, it was very probably a naive comment about the power of as-then-understood computer technology's capability to replace manual labor or work. In the 1990s, it was very probably a naive comment about the state of the art of AI at the time.

Now, the remark is about actual inexpensive robotics in the context of what seems to be a genuine AI and computer vision and control renaissance, funded by government and corporations. Those are all just different remarks. To package them all up as one "thing that people have been saying" does a real disservice to the discussion. And it's a pose.

In the current case, I find it difficult to say whether the remark will turn out to be naive. However, any competent discussion of whether or not it is has to rely on an assessment of the current situation of computer vision, AI, control theory, economics, and law.

Jobs are only being created where human labor is still cheaper than robots. But the price of human labor keeps rising, so the factories continually keep moving to new countries with cheap labor. Meanwhile robots just keep getting cheaper. Inevitably there must come a point where there isn't anywhere to move the factories for cheap labor, as the whole world has come out of extreme poverty, while robots have gotten cheaper than what any human can offer their labor for.