On the surface-level, yes, it's semi-convincing. On the other hand, it lacks all the nuance of a human being.
Ask it `Give me a taco recipe in the style of Stanley Unwin` and you'll get something that has the style, something that's superficially Unwin-like, but lacks the contextual nuance of the humour completely. In fact, I'd say ChatGPT doesn't do humour, especially nuanced humour that's in the delivery and not in the punchline itself. Humour is a very abstract thing and there's so many styles of it. Can it tell you a joke? Yes, but that's not humour in its entirety.
Just an observation, but the LLM isn't expressing itself, it's words for the sake of words because they predict the next word in a sequence. It's like saying "what's the big deal about Allen Ginsberg? You drop a few f-bombs and a bit of flowery language, that's Ginsberg in a nutshell", whilst completely ignoring the expressivity of self.
What makes writing as a medium so interesting is the conveyance of thought from one human being to another. Is an LLM interesting because it can sometimes be a little alien whilst still having some human-like quality? Sure. But an LLM is just an LLM, it doesn't have experiences or novel new thoughts that express a larger ideology. As a human being, I unfortunately just don't relate to LLMs, even if they are a reflection of the content they're trained on, which is ultimately human. It just lacks soul.
If the LLM can't experience the zeitgeist in the way that we do, it can't really capture such a big chunk of what human self-expression is.
Anecdotal, but I find the average HN exchange to be far more interesting than anything an LLM has ever output to me in terms of conveying meaningful ideas. I've tried debating with ChatGPT itself a few times, that's been—hit-and-miss.