ABSTRACT
Technology has been supplanting human labour since the invention of the lever and the wheel. That it did so did not pose a social problem until the Industrial Revolution and the shift to waged labour as a fundamental economic institution constitutive of society. In the United States, fear for economic and social dislocation proceeded in two waves: in the 1920s and 1930s, persistent mass unemployment resulting from “technological unemployment” was feared. It did not happen. In the 1950s and 1960s, the same fear recurred, this time grounded in automation as the spectre of “structural unemployment”. It didn’t happen. The fear has returned today, instigated by robotisation and artificial intelligence. The professional consensus is that now, as before, the fear of persistent mass unemployment is ill-grounded. Unemployment is a consequence of low growth, not technology: as before, new jobs will be created as others are destroyed. But the professional consensus today is that those new jobs will be bimodally distributed into non-routine, well paid, highly skilled work and non-routine low paid semi-skilled work. Technology is relatively well-paid but routinised work, work amenable to technological displacement. The prospect then is of widening and deepening income inequality. This poses a problem for public policy.
