Louis Rossmann vs AI Slop: The Manifesto for Writing Like Humans

The problem is simple: nowadays anything written by AI tends to sound identical to everything else. Phrases like "it is important to note that" or "let's explore together" have become the trademark of ChatGPT garbage and similar tools. Rossmann decided to create a practical guide, designed for Claude (Anthropic's AI), that instead of producing flat and generic text teaches how to write like a real person.

The rules aren't overly complicated, but they're strict: ironic and direct tone, short sentences with varied rhythm, concrete examples, and most importantly zero pointless jargon. It's like telling a machine "don't write like a machine, write to me like you're sitting across from me at a bar". No disclaimers like "as an AI model", no arbitrary listicles, no text that looks extracted from a university textbook.

What makes this project interesting is the fact that Rossmann doesn't go full anti-AI crusader. He's not saying "AI is evil". He's saying: "if you use AI, use it to do smart things, not to spam the web with content that all looks like it came off the same factory conveyor belt". It's a critique of misuse, not the technology itself.

The CLAUDE.md file is basically a portable style manual: you can copy-paste it in front of Claude (or any other model) and the text that comes out has at least a fighting chance of being readable as if written by someone with an actual personality. It's not a guarantee, but it's a concrete first step.

Basically, Rossmann is trying to teach artificial intelligences how to not behave like artificial intelligences. Sounds almost paradoxical, right? But that's exactly the point: if an AI tool is supposed to be useful for creating real content, it needs to learn to mimic how a self-aware human being speaks, not a randomly compiled encyclopedia.