There are distinctions and comments to be made.

The first is the difference between the practical and the theoretical. Economic activity is generally practical and productive in the sense that we don't do these things for themselves, but for something else. However, an infinite regress of practicality is impossible, because the practical cannot justify itself. It is always for the sake of something else, so you must ultimately arrive at something done for its own sake.

The second is the distinction between the internal and external effects of work. When people talk about AI taking jobs, they focus almost exclusively on the external effects (the actual resulting products and services). They rarely touch on the effect of work on the worker. Properly dignified work[0] directed toward proper ends is an act of love and service that also develops and perfects the worker as a human being.

A third is the distinction between the pathological condition by which a person seeks vacuous simulacra and stimulation and in a manner that causes them to fail to develop socially, and the claim that such stimulation and simulacra can actually replace healthy social relationships. The latter is not true, the former already exists to some extent.

I would also reject "economist" framing of human relationships as such and of human persons.

And finally, the Singularity as bona fide intelligence is nonsense, but one might argue that, at least in principle, some kind of hyper-developed and virtually unbounded instrumentation of computation ("AI") is possible. But AI lacks inherent ends and could not possibly possess its own ends (because it is not a true integral whole and thus lacks immanent causality), and so it would be an instrument of human will. Whose will is the question.

[0] Hence, why I think that while using AI to replace human reason is bad, the reduction of tedious and mind-numbing work is not.