ai;dr

Readers revolt: AI for code is fine — but hands off our words

TLDR: Author says AI is great for coding but drains the life from human writing, sparking a debate on effort and authenticity. Commenters split between co-writing with AI and demanding “proof-of-work” prose, with horror stories of chatbot fluff and a warning that the writer–reader social contract is at risk.

A fiery post just drew a line in the silicon sand: AI-generated code = progress; AI-written articles = soulless slop. The author says writing is a window into a person’s mind, and outsourcing that to a chatbot makes the “dead internet” theory (the idea that bots now drown out real people) feel a little too real. Spicy twist: typos now read as proof you’re human. Cue the crowd gasping.

The comments lit up. Team Nuance showed up first, with alontorres arguing that if there’s a long, hands-on back-and-forth with the model, the result can be legit, not lazy. The jokes flew too: soperj dunked on a stray “specially” with a deadpan “how we can tell it wasn’t written by an LLM,” and elischleifer called the “messier = better” line a bit much. The most upvoted anxiety? numbers’ tale of a boss whose AI-polished emails got so fluffy that the real message was buried “somewhere in the middle.” Meanwhile, losvedir dropped Oxide’s take: AI prose breaks the social contract—writing should usually take more effort than reading. The thread split between “proof-of-work prose” diehards and AI-as-coauthor pragmatists, with everyone agreeing on one thing: transparency matters, because nobody wants to decode corporate fluff wrapped in robot sparkle.

Key Points

  • The author views writing as a direct lens into human thought and values human-authored prose for its intention.
  • They use LLMs extensively for coding tasks and find them efficient, citing documentation, tests, and scaffolding as examples.
  • The author believes AI-generated articles feel low-effort and contribute to concerns about the authenticity of online content.
  • Signals like typos and imperfect grammar, once negative, now appear to indicate human authorship to the author.
  • The author acknowledges these human-like signals can be imitated, questioning their reliability as indicators of authenticity.

Hottest takes

"long process of back-and-forth with the model... I don't think there's anything wrong with that." — alontorres
"How we can tell that this wasn't written by an LLM." — soperj
"there's so much ChatGPT fluff around it." — numbers
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.