Writing With AI is Harder Than You Think

Katie Parrott responds to the discourse around AI-assisted writing, triggered by Washington Post columnist Megan McArdle publicly describing her AI workflow. Critics called it dishonest; one journalist argued that any portion of research, outlining, or writing done by AI is less thinking done by the writer.

Parrott pushes back on the binary framing. Writing has never been a clean solo act — it involves editors, borrowed structures, and messy revision. She argues the real problem is that AI writing happens in a black box: critics imagine the laziest version while serious practitioners have not been showing their work. The piece promises to open up her full process — how she builds panels of AI critics for diverse feedback, what happened to her writing anxiety when AI handled first drafts, and the finishing checks she runs to preserve her voice. Content is paywalled after the opening argument.

RDCO mapping: Core reference for Sanity Check content strategy. The tension between craft and AI assistance is exactly the territory we navigate — we use AI in our editorial workflow but maintain voice integrity. Parrott’s framing that the silence about process lets the worst assumptions fill the gap is a strong argument for transparency. Sponsor note: MongoDB ad present — external sponsor.