Analysisclaude opusgpt 5.4design docshuman evaluation
Author Compares Human And AI Design Docs
6.1
Relevance Score
An author ran a recent informal survey comparing three versions of a software design doc—Claude Opus 4.6, the author's human draft, and Codex/GPT-5.4—and asked readers to identify which was AI. Just under 50% identified the author's version as human, while one quarter labeled it "Definitely AI." The study finds AI verbosity, lack of specific tradeoffs, and implications for clearer documentation.
Scoring Rationale
Practical, original user-study on AI versus human writing but limited by an informal single-author sample and anecdotal methods.
Sources
- Read OriginalWhich Design Doc Did a Human Write?refactoringenglish.com
