To parrot arguers claim that these system just regurgitate the training data set and give you the most plausible next word, and they claim that's bad/useless.
The anti-parrot arguers claim that's basically how humans work too (even the smarter ones), and that just because ChatGPT is a "parrot" doesn't necessarily mean it cannot be as good or better than humans. It might not be entirely there yet, but it's just a matter of time it becomes a good enough parrot to be indistinguishable/better than humans.
A parrot can be a good system if the problem is "Collect and sort the available data about X (and provide reference)".
A parrot is a disaster if you ask it "What do you think of X": a parrot does not think - unless you make it reason, verify and criticize its ontology, and at that point it is useless that we call it a "parrot".
Why need it be a disaster?
If its answer is useful, why bother about whether the answer was the result of thought or some other process?
Most opinions you will hear from people are also just regurgitated amalgamation of things they have previously heard or read on the subject. Also, in general, being able to aggregate and summarize into words what hundreds of people thing about X seems like probably an even more useful tool than getting one person's thought, unless it's something no one has ever had any thought on before.
Even then, I don't think it would be outside the scope of possibility for the parrot to find things that are similar to X, find opinions on those, and somehow transpose those opinions onto X.
Which humans? There may be a correlation between people who are capable of watching Netflix quality series for 4-5 seasons of more of the same and admirers of ChatGPT.
Basically if "content" (as opposed to "art" or attempts at art) does not bother you, ChatGPT does a good enough job already.
On the other hand, the most vocal opponents imagine themselves in a world where not 90% but 99.999% is blogspam and Netflix. Making entertainment worth watching impossible to find.
And I stated that I cannot think of a human job in which intelligence is not required, which cannot be replaced by deterministic automation.
The fact that many are mentally incompetent cannot be a reason to produce something similar. When somebody does a job, we expect such job to be done properly, and "brains" are pretty much required, intelligence, if only for the occasional cases outside norm - otherwise, there is automation.
> if its useful
It is not easy to confine an affabulator to the contexts where it will be useful.