The Decline and Fall of Technological Unemployment

The concept of "technological unemployment" appears to have peaked in 1936, after which it has been on a steady decline. Meanwhile "Luddites" -- which I take to indicate mainly disparaging mentions of Luddites has been on the rise, slowly and unsteadily throughout the 20th century and into the first decade of the 21st.

I'm working on a long post but in the meanwhile here's a thought: maybe technology "creates more jobs than it destroys" because it isn't really "labor-saving" in the first place. That is to say, machines aren't intrinsically productive but simply mask the uneven exchanges that occur between the industrialized global North and the extractive South.

It would admittedly be difficult to prove such a hypothesis but my question is: what evidence is there for the commonly-assumed believe that machines are "intrinsically productive"? I know, I know, there doesn't need to be any evidence for the assumption because it is "self evident" to the extent that anyone who questions that self-evidence reveals themselves to be a fool, a crank, a crackpot, a Luddite and an enemy of progress.