
What happened
Thomas Ptacek published advice on writing with LLMs, stating "Rule Number One: You may not use a single word an LLM suggests to you," calling it "intellectual personal protective equipment" and urging readers to be strict.
Why it matters
The rule responds to text that "has that weird smell to it," and is also a good principle to help stay disciplined.
What to watch
Thomas later shows a screenshot of his personal copyediting tool and provides a prompt to help kickstart building your own. The test is whether readers stick to it.
WHO IT HITSWriters, editors, and blog authors who use LLMs to check or polish their prose now have a concrete rule to debate: treating LLM-suggested phrasing as off-limits, while still using fact-checking, spelling and grammar tools.
Summaries like this, every morning.
Thomas Ptacek's rule sits at the center of a broader piece about how to use LLMs in writing. The article positions LLMs as copyeditors rather than writing assistants, a distinction that runs through the examples. The rule to never use a turn of phrase suggested by an LLM is described as feeling good, and the text's "weird smell" is given as one reason. The article notes it is also a good principle to help stay disciplined.
Simon Willison, whose blog hosts the piece, echoes the sentiment. He writes that he won't let LLMs write content for his blog, but uses them for fact-checking, spelling and grammar, and as an occasional thesaurus. The article links to his proofreading prompt. The piece later shows Thomas's personal LLM copyediting tool and provides a prompt to help kickstart building your own.
The stakes appear to hinge on whether writers can hold a strict line against LLM-suggested wording while still using LLMs for the tasks the article permits. For blog authors and editors weighing how much of their prose to run through an LLM, the piece offers a concrete rule to test rather than a general argument about AI writing.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Atomicwork launched an AI Coworker Marketplace with certified AI coworkers for cloud infrastructure, access ma…
FrigadeHQ's Hob installs on Windows, Mac and Linux, scans the PC, and offers an AI model suited to its memory…

Amazon is adding a texting feature to Alexa for Shopping, letting US customers describe or photograph items an…

Carnot AI Inc, which develops and provides the enterprise AI platform Jinba, was accepted into NVIDIA Inceptio…

In a guest commentary, R K Anand of Tensordyne argued challengers should target inference—not training—citing…

Thinking Machines released Inkling 2 Small, a 594B-total / 17B-active multimodal MoE, available now on Tinker
