Thomas Ptacek has shared his approach to using large language models during the writing process, establishing a firm boundary between drafting text and editing. According to reports highlighted by Simon Willison on 4LUP.COM, the core rule of this methodology is that writers must never use a single word or specific turn of phrase suggested by an LLM.
The Copyeditor Approach
While generative artificial intelligence tools are frequently marketed as complete writing assistants capable of drafting whole essays or blog posts, this perspective restricts their use strictly to proofreading functions. Ptacek reportedly uses LLMs solely for tasks such as fact-checking, checking spelling and grammar, and acting as a thesaurus. Treating the AI as a strict copyeditor rather than a creative partner aims to protect personal voice and intellectual independence during content creation. The announcement did not specify the exact technical architecture or the specific model brands utilized in this workflow.
Open Questions And Unknowns
Several details regarding this writing framework remain absent from the available information. The announcement did not specify the exact proofreading prompts used to elicit thesaurus suggestions or grammar checks without triggering unwanted creative completions. It also remains unclear how writers should evaluate the risk of stylistic bleed, where exposure to LLM-generated phrasing might subtly influence human word choice even when direct copying is avoided. Observers will need to watch how independent writers balance efficiency gains from AI tools against the preservation of authentic voice.
Adopting strict personal protective boundaries around generative writing tools may offer a sustainable path forward for independent bloggers.
Lead image: Luis Quintero / Pexels
