This elides three distinct effects: Amazon is growing at the same time as they’re roboticising — this confounds robots with growth. They are consuming the retail business, which confounds their hires with lay offs at their competitors. The robot question is buried in there. Warehouse AI will see those entire facilities go dark (robots don’t need lights on). Perhaps AMZN will have other work for 1e9 people, but it seems unlikely that robotics leads to more hiring, and much more likely it leads to 90% reductions in any work where behavior follows a pattern that can be replicated, and no human customer is present to enjoy (and demand) social interaction.
So the question becomes, can we do so much more that the so-many fewer people involved in each element of more out number the losses from roboticisation. This clip from an address to the fire fighters association this year covers similar ground, arguing that most businesses see machines as a way to cut labour costs, and therefore (because they are not suffciently fly-wheel like) and, therefore, that the government should probably just create a facade of public safety and declare driverless vehicles illegal at least until the fly wheel spins up :-) https://youtu.be/t2dct9ErA_g?t=27m