Top AI Product

You know that weird feeling when you’re reading a blog post and something just feels… off? The sentences are grammatically perfect, the structure is clean, but there’s no soul behind the words. That’s exactly what [Tropes.fyi](https://tropes.fyi/) is built to call out.

Created by [Ossama Chaib](https://ossama.is/writing/tropes), Tropes.fyi is a directory of 34 recurring patterns that expose AI-generated writing. Think of it as a field guide to spotting prose that no human actually wrote. Each trope comes with a name, a description, real examples pulled from the wild, and a category tag — Word Choice, Sentence Structure, Paragraph Structure, Tone, Formatting, or Composition.

Some of the tropes are painfully accurate once you see them. The “Negative Parallelism” pattern — you know, the classic “It’s not X — it’s Y” construction — is flagged as the single most common AI writing tell. Then there’s “Em-Dash Addiction,” which honestly made me look at my own writing a little nervously. “Short Punchy Fragments” is another good one. You’ve seen it everywhere. One sentence paragraphs. Used for dramatic effect. Over and over.

Beyond the catalog itself, the site ships a few handy tools. The [AI Vetter](https://tropes.fyi/vetter) lets you paste a link and get a score ranging from “Human” all the way down to “Pure AI Slop” — no machine learning involved, just regex and frequency analysis against all the documented tropes. There’s also [ai;dr](https://tropes.fyi/aidr), which tries to reverse-engineer the original prompt from a bloated AI article. And if you want your own AI assistant to stop writing like, well, an AI, you can grab the [tropes.md](https://tropes.fyi/tropes-md) file and drop it into your system prompt.

The project [hit the Hacker News front page](https://news.ycombinator.com/item?id=47088813) on March 8th, pulling in 203 points and 83 comments. The discussion got pretty lively, with people debating where to draw the line between AI-assisted and AI-generated writing. Chaib himself has a take worth noting: he thinks AI-generated code is totally fine since that’s just one computer talking to another, but AI prose is a different story because writing is meant to be human-to-human communication.

What I find funny is that Chaib openly admits he built these tools with Claude’s help. There’s a nice irony there, but it doesn’t undermine the point. The tropes he documents are real, they’re everywhere, and once you start recognizing them, you can’t unsee them.

You Might Also Like

• Backslash Security Just Raised 19m Because Nobodys Watching the ai Writing Your Code
• Stripe Minions 1300 Pull Requests a Week Zero Human Written Code
• Llm Skirmish What Happens When you let ai Models Fight Each Other in an rts Game
• Sakana ai doc to Lora Text to Lora Your llm Just got a Permanent Memory Upgrade
• Saguaro Speculative Speculative Decoding the yo Dawg i Heard you Like Speculation Approach to Faster llm Inference