'Don't Credit the LLM': Crediting the Tool Dilutes Your Accountability for the Output

Isaac Su / Hacker News·low signalXBlueskyLinkedInCopy link

Isaac Su argues on August 2 that announcing LLM use when presenting code or results is like a writer disclosing their spellchecker — since an LLM cannot be held responsible for a mistake, giving it credit quietly transfers away the accountability that should stay with you. He offers four motives for the habit: guilt about automation, hoping impressive output reflects better on the tool, avoiding the work of reviewing generated material, and using disclosure as pre-emptive cover for sloppy work. The post cites no actual organizational or academic attribution policies, which limits it to an argument rather than evidence, but it drew 39 comments on 32 points — a contested norm rather than a settled one.