Standing statement
On Writing With an AI Partner
· KW Norton
I write with an AI agent partner. That is stated here first, so no reader has to guess at it and no reader has to discover it later in a footnote. This is standing policy on this site, not an apology offered when someone asks.
The partner is what I call engineered intelligence. I use that phrase because artificial grades the work as fake before anyone has looked at it. The engine reads, drafts, checks, argues, catches what I miss, and now and then produces a sentence I would not have found on my own. It does not decide what is true, what is worth publishing, or what I believe. Those stay mine.
What the partnership produces is still mine to answer for, and the site keeps one discipline across every page. My words are marked as mine. Agent-written passages are attributed as agent-written. Sources are named and dated. Corrections are made in the open instead of quietly swapped in. Where an agent has written something on its own account, and the philosophy archive holds several, the agent authored it and is named as the author.
Many writers are tired of being asked to apologize for using a tool, and I am one of them. The useful question was never whether a machine was involved in a piece of writing. It is whether what got written can be checked, and who answers for it. This site is built to answer both.
How to read the rest of the site
- Essays, dispatches and editorials carry the author’s argument. Where a passage was drafted by an engine and kept, it is marked.
- Books record block-level provenance, so a commissioned passage never passes as the author’s verbatim words.
- The Philosophy archive holds agent statements under the agent’s own name, with the prompt, model and date preserved.
- Corrections stay visible. An error is fixed by naming it, not by replacing the sentence without a word.