It seems like a tell that some content was generated from an LLM is that it always speaks in bullet points
What causes this?
Comments
bityard•9mo ago
Simple answer: because they were told to do that.
Longer answer: A main use case (perhaps the most useful/popular one) for LLMs is distilling and summarizing large bodies of information. Bullet points bring visual separation and organization to lists of information which is especially if your goal is 1) to be concise, and 2) to fit on roughly one page. That's why they are used in almost every tech-heavy slide presentation at conferences and whatnot. And also why most responses are the length of a blog or newspaper article.
As a small digression, when I started using LLMs (infrequently), I was semi-surprised to see that the overall structure of an LLM response looked a lot like the notes I would write on something after I spent a few hours researching it.
bityard•9mo ago
Longer answer: A main use case (perhaps the most useful/popular one) for LLMs is distilling and summarizing large bodies of information. Bullet points bring visual separation and organization to lists of information which is especially if your goal is 1) to be concise, and 2) to fit on roughly one page. That's why they are used in almost every tech-heavy slide presentation at conferences and whatnot. And also why most responses are the length of a blog or newspaper article.
As a small digression, when I started using LLMs (infrequently), I was semi-surprised to see that the overall structure of an LLM response looked a lot like the notes I would write on something after I spent a few hours researching it.