Isaac Su argues against the growing habit of crediting LLM tools for work output. He compares it to writers disclosing spellcheckers or pilots announcing avionics versions—unnecessary and odd. When peers cite an LLM for support request counts or admit an LLM wrote unit tests, Su asks why the disclosure is needed.

Su identifies four motivations: believing credit is due, feeling like an imposter, being impressed by LLM performance, or outsourcing work without review. He notes the work may be only 80 percent complete with no one knowing which 20 percent is missing. Su dismisses most reasons but challenges the first: credit and accountability are inseparable.

A tool can only claim credit equal to its accountability. Crediting an LLM dilutes human responsibility. If work is profound, take full credit; if flawed, own it and improve. This principle predates and will outlast LLMs. The Hacker News post sparked an HN discussion14 on the topic.