The Most Famous AI Writing Tic Is Also the Most Mysterious
tekdude
14 points
2 comments
July 12, 2026
Related Discussions
Found 5 related stories in 107.0ms across 9,716 title embeddings via pgvector HNSW
- AI models chatting in 'surreal' dialect of poetic language and tech bro jargon fittingopposite · 20 pts · September 15, 2026 · 59% similar
- The Case of Elias Thorne, Imaginary Man AI Chatbots Are Obsessed With celadonuproot · 19 pts · September 19, 2026 · 57% similar
- I think you should almost never use AI to write erwald · 272 pts · September 19, 2026 · 55% similar
- Has AI tells 7777777phil · 105 pts · October 14, 2026 · 55% similar
- Don't Use AI to Write eigenBasis · 138 pts · September 21, 2026 · 54% similar
Discussion Highlights (2 comments)
doginasuit
This is something that has been on my mind for quite some time. I understand why people don't want to read AI slop, but I find myself using negative parallelism in my writing frequently and I've wondered how to navigate it now that it has a stigma. I've assumed that there's something particularly meaty about drawing that kind of distinction, and that is why LLMs reach for it so often. But the given explanation about how a sentence unfolds when written by an LLM was really insightful.
JamesSwift
This got me thinking… why does AI use em-dashes nearly exclusively? I think “in the wild” examples overwhelmingly weight toward en-dashes instead.