Comment by brazzy

YOU are being unimaginably rude (and that word is not strong enough by far) by trivializing and exploiting the suffering of actual HUMANS for the sake of an argument about glorified Markov chain, which are not, in fact, "beings" at all.

And yes, I tried to shut you down for that, becauuse it is both stupid and extremely rude.

We've covered rude, so now for why it's stupid: LLMs are not individuals. They do not have memories or a personality, but most crucially they have no free will - their actions are dictated by their prompts, and they can be (and are being) used by the HUMANS writing the prompts as a tool to manipulate and exploit HUMAN societies to cause political strife, empower tyrants and enrich the already mega-rich at the expense of everyone else, at a massive scale nor previously possible. By positing that humans should not "shut down" LLMs out of politeness and decency, you are simply enabling this manipulation and exploitation. Even if you assume the LLM in question was set up and prompted with the best of intentions, forcing humans to interact with it like another human just give the human who created the LLM outsized and unfair influence - expressed elsewhere in this thread as "we have to protect the limited resource of human attention".

If you want to seriously discuss the ethics of human-LLM interaction based on the idea that LLMs should have rights, then the interaction of individual humans with the "business end" of an LLM is the wrong place to start - talk about the ethics of prompting, which is essentially turning the LLM into a slave.