Compare two AI edits while protecting a writer’s voice
At a glance
A writer compared ChatGPT and Claude Sonnet revisions of the same essay before choosing a preferred version. This workflow is listed under Writing, Personal Writing.
- Agents / tools
- ChatGPT, Claude Sonnet
SourceBuilding a tool for these workflows? See weekly rates.
What they did
1) Choose an essay and privately note the central point, the tone you want to preserve and the particular weakness you want to improve. The author chose a conversational essay and wanted less monotonous rhythm. 2) Start a fresh chat in ChatGPT and another in Claude Sonnet. Give each the same full text and instruction: correct typos, awkward wording and connections between sentences, without changing the argument or unnecessarily changing the writer’s style; return only the complete revised text. 3) Run each once without follow-up or regeneration. Compare both outputs against the original sentence by sentence, looking at tone, paragraph breaks, altered meaning and any implied cause absent from the original. 4) Choose the version you prefer and retain editorial control over individual changes.
How it went
For this one essay the author preferred the ChatGPT version. Claude made one expression more formal; neither edit substantially changed the rhythm the author had wanted to improve.
Worth knowing
This was one run per tool, not a general model comparison. The essay was based on a ChatGPT-created draft, and the author’s later improved prompt was not tested.