I've been running IRC markov chain bots for ~20 years. In the last couple years I've been running local LLM alongside them. While there are people that still prefer the markov chain bots the majority invoke the LLMs. But maybe that's because I picked the most hallucinate'y, least refusing, most funny models I can (like mistral-7B fine tunes) instead of something smart, locked-down, and low temperature like chatgpt's LLM as a service.

If anything seeing the LLM and markov bots side by side has really reinforced how much of the markov bot "humor" is human perception imposed on chance outputs. The markov's "learning" ability is still far superior though.