You can find an estimate for the impact of automation on jobs to support more or less any prior opinion you happen to hold. Apparently sensible forecasts from apparently respectable organisations range from the benign to the apocalyptic. One reason why it can be hard to make sense of that is that different factors – some potentially with opposite effects – can get bundled together into an agglomeration of technologies in an agglomeration of sectors and contexts having an agglomeration of effects. Disentangling any of that then becomes next to impossible
This useful short post separates out four different forms of automation, briefly explaining the employment impact of each. That doesn’t magically give us a single number which is somehow better than all the other single numbers. But it does provide a framework for understanding and debating all the other single numbers.