My take is that the generated stuff is terrible for communications.
It feels great to use, direct your machine minion to fill out your thoughts for you, but holy hell does it suck to be on the receiving end. Least of all is the disrespect, they don't care enough to even talk to you but worse is having to try and reason through that big incoherent blob.
Probably to only reasonable thing to do is to try and get your own mechanical agents to produce summaries. Inventing the lossy expansion algorithm(like compression but things get bigger on the wire), And we wept.
Now I am all depressed because it is probably inevitable, apparently thinking is hard and in general people are all to happy to outsource it to the machines.
But even that is grossly inefficient.
Let's say I'm making an HN comment. I have a one-sentence idea, I get an LLM to expand it into an impressive-looking (or oppressive-looking) wall of text, and then I post that. Well, let's say 10 people see it. And each one of them has to either plow through it on their own, or paste it into LLM to get the summary.
But even with one-to-one communication, it's still terrible, as you say. You can't be bothered to clarify your idea, but you're trying to use an LLM to make up for your lack of thought? So you're going to make me plow through that huge blob of text to try to understand what your thought was, the thought that you couldn't bother to actually really think through. That's far less efficient than you, the sender, actually doing the thinking.
But it lets the sender be lazy. And the sender is the one in control.