Hacker News Reader

I think I must be an AI. I’ve written those types of phrases a lot long before the advent of LLMs and I love em dashes. LLMs were trained on human text. Where do you think they got these patterns from? Looking at the rest of the text, I don’t see some of the other LLM tells like excessive usage of bullet lists, and repetition of the same ideas over and over using different language.

It’s a well written, engaging article. Why would it need a disclaimer if they used AI on part of it? They’re not arguing against the usage of LLMs.

Seeing these constructs in a text is not just a vague hint that it was AI-generated. It's a smoking gun.

It almost never does.