The article argues that natural-language text cannot be losslessly transformed, as any rewrite or rephrase alters meaning and risks losing the author's original intent. It highlights Sophie Alpert's guideline that engineers must stand behind every sentence produced with AI assistance, rejecting the excuse that AI generated the content.
Background
This post expands on ethical guidelines for using LLMs in technical writing, focusing on the risk of semantic drift when AI rephrases human-authored content.
- Source
- Simon Willison
- Published
- Aug 12, 2026 at 07:48 AM
- Score
- 5.0 / 10