A Hacker News thread titled “Don’t just paste the AI at me” calls out the habit of dumping raw LLM output into replies daily. When someone answers with a preamble such as “Here’s what Claude said:” and then pastes eight hundred words of unedited ChatGPT text, the author argues it reveals a lack of personal effort. The post likens the behavior to forwarding a generic email.

The critique stresses that askers seek the responder’s judgment, not a boilerplate summary. AI models exude confidence even when half the content is vague or incorrect, so copying them signals “I didn’t read your question.” The author advises scanning the model’s bullets, keeping what checks out, and rewriting everything else in one’s own voice before sending it back.

Ultimately the piece treats LLM output as a junior draft, not a finished answer. Readers are urged to edit, cut, or discard the suggestion and, if nothing adds value, simply reply “I have nothing to add.” This discipline preserves credibility and keeps technical discussions focused on human insight rather than recycled AI prose in community forums.