The link goes to a post from 2012, in which he works through a simple example to show that technological progress which raises the efficiency of capital-intensive production techniques can raise the return to capital and lower the return to labour: leaving workers worse off.
Right, so, does my story go beyond this?
In terms of the productivity paradox, the additional step is to ask what happens when a switch toward more capital-intensive techniques happens in some parts of the economy, reducing wages. My answer is that falling wages lead to the creation of some new, low-productivity employment in some industries. They also deter a switch from labour-intensive to capital-intensive production in others. If wages fall enough and there is a high fixed cost to changing techniques (because firms have to learn how to make highly automated production facilities work, for example, or need to have a lot of data and do a lot of experimenting before their machine intelligence is able to do better than human workers) then technological progress which ought to encourage a shift toward more capital-intensive production techniques can quickly fizzle out.
Basically: wages affect the choice of technique, and the choice of technique affects both the level and the growth rate of productivity (because if you don’t start using AI, for instance, then you don’t get the knock-on improvements that allow it to be deployed effectively in more contexts). So if technological progress causes wages to fall, that not only harms workers: it can also help usher the economy into a low productivity growth trap.
There’s also a lot more to the book, which looks at demand-side effects, the effects of a scarcity of intangible (or social) capital, and much more. But this is the point I wanted to get across in the productivity paradox post: we are underusing technology because workers are too cheap.