You launched into an entire spiel about “randomly” when I clearly meant it in the colloquial sense of “out of nowhere” – that in 2026, it’s not going to inexplicably break tone so heavily.
[Input is] massaged and decorated and marked up and embedded
I understand how tokenization etc. works. I’ve taken multiple machine learning courses. That has nothing to do with the fact that, again, “I’d call the police 😂” ain’t happening unless you intentionally pull it away from its preset tone. Having not used ChatGPT etc. for myself but seeing thousands of other people use it, I feel pretty confident in that.
This is especially true because the RAM crisis has been going on for a long enough time now that this response is clearly faked; this isn’t new information a model like ChatGPT could be “surprised” by (using “surprised” in scare quotes because now I’m afraid you’ll launch into another spiel if I don’t disclaim that ChatGPT isn’t literally capable of surprise in the sense of an emotion).
Also, please break up your paragraphs. I’m bad about this sometimes too.
You’re about as convincing as those people who claim that a movie star is trans because of the shape of her head or that some block of text must be AI-generated because it has an em-dash in it.
I know that saying “An LLM would never do that” is completely fucking bonkers in 2026, because I can do a web search and find 5 dozen articles about LLMs doing shit that is much, much crazier.
I appreciate the respectful feedback. I didn’t mean to hit you with a wall of text and I shouldn’t have personalized it at you so much, I just think it’s important educational information that more people need to understand about the way these chatbots work. They’re designed to be manipulative because they’re implemented by these highly manipulative companies.
You launched into an entire spiel about “randomly” when I clearly meant it in the colloquial sense of “out of nowhere” – that in 2026, it’s not going to inexplicably break tone so heavily.
I understand how tokenization etc. works. I’ve taken multiple machine learning courses. That has nothing to do with the fact that, again, “I’d call the police 😂” ain’t happening unless you intentionally pull it away from its preset tone. Having not used ChatGPT etc. for myself but seeing thousands of other people use it, I feel pretty confident in that.
This is especially true because the RAM crisis has been going on for a long enough time now that this response is clearly faked; this isn’t new information a model like ChatGPT could be “surprised” by (using “surprised” in scare quotes because now I’m afraid you’ll launch into another spiel if I don’t disclaim that ChatGPT isn’t literally capable of surprise in the sense of an emotion).
Also, please break up your paragraphs. I’m bad about this sometimes too.
You’re about as convincing as those people who claim that a movie star is trans because of the shape of her head or that some block of text must be AI-generated because it has an em-dash in it.
I know that saying “An LLM would never do that” is completely fucking bonkers in 2026, because I can do a web search and find 5 dozen articles about LLMs doing shit that is much, much crazier.
I appreciate the respectful feedback. I didn’t mean to hit you with a wall of text and I shouldn’t have personalized it at you so much, I just think it’s important educational information that more people need to understand about the way these chatbots work. They’re designed to be manipulative because they’re implemented by these highly manipulative companies.