My primary worry since the start has been not that it would "replace workers", but that it can destroy value of entire sectors. Think of resume-sending. Once both sides are automated, the practice is actually superfluous. The concept of "posting" and "applying" to jobs has to go. So any infrastructure supporting it has to go. At no point did it successfully "do a job", but the injury to the signal to noise ratio wipe…
The general tone of this study seems to be "It's 1995, and this thing called the Internet has not made TV obsolete"; same for the Acemoglu piece linked elsewhere in the. Well, no, it doesn't work like that, it first comes for your Blockbuster, your local shops and newspaper and so on, and transforms those middle class jobs vulnerable to automation into minimum wages in some Amazon warehouse. Similarly, AI won't come for lawyers and programmers first, even if some fear it.
The overarching theme is that the benefits of automation flow to those who have the bleeding edge technological capital. Historically, labor has managed to close the gap, especially trough public education; it remains to be seen if this process can continue, since eventually we're bound to hit the "hardware" limits of our wetware, whereas automation continues to accelerate.
So at some point, if the economic paradigm is not changed, human capital loses and the owners of the technological capital transition into feudal lords.