while this may be novel, since its a relatively new phenomenon, it is not what makes poetry beautiful and compelling to actual humans, which is one human soul communicating to another.
sure, the unsophisticated and uneducated might be amused, but theres nothing to connect to when theres only a cold hard emotionless machine on the other end.
I've been really frustrated by the "no understanding" or "stochastic parrot" attitude about these language models, because my own experimentation convinced me they are wrong about that. This resonates with the feelings I have about that.
However, I think it's worth reconciling the parent's point with your own worldview. You may feel attached to these LLMs, but that's mostly because they're designed as affirmation boxes. They fill-in-the-blank with whichever token the sampler passes to them, it really is that simple. Humans and AI can arrive at the same outcome (circumstantially), but they will use different means to go about it. ChatGPT and it's contemporary LLMs do not have a model of self. They're good at convincing you otherwise, and it's completely fine to ascribe human emotion to that writing - but you should at least acknowledge that it's just writing.