Wikipedia:Signs of AI writing
11 points by Corbin
11 points by Corbin
Folks have repeatedly wondered how it's possible to recognize that a human did not write a blog post. Here's a listing of tips from Wikipedians. Many of these have also come up in other wiki communities; personally, I recognize many of these from the esoteric-languages wiki, which also uses Mediawiki.
A useful resource! On the other side of the coin, I recall there were also folks piling on people who said they couldn't discern LLM-generated text; perhaps the relevant section in that page will reassure them:
Do not rely too much on your own judgment. Humans are notoriously bad at distinguishing human and LLM-generated text. While research on humans' abilities to detect AI-generated text is still limited, a 2025 study has shown that human ability to distinguish LLM text from human is no better than random chance.[4] Another 2025 study on German theses has shown that humans managed a "recognition rate of 57 % for AI texts and 64 % for human-generated texts".[5]
A 2025 preprint has shown that heavy users of LLMs can correctly determine whether an article was generated by AI about 90% of the time, which means that if you are an expert user of LLMs and you tag 10 pages as being AI-generated, you've probably made one false positive. People who don't use LLMs much do only slightly better than random chance (in both directions).
I wonder what happens if you give your LLM this article and ask it to write better.
Hard prompts and contexts are like polynomially-strong steering. Reinforcement learning is like exponentially-strong steering. The reinforcement-learned behaviors, which include most of the quirks and tells documented in the article, tend to dominate.