Claim
AI produces identical voice patterns for every writer, so it cannot ghostwrite. It can function as a great editor, suppressing its own patterns to improve structure and accuracy without erasing the author's voice.
Mechanism
A ghostwriter harmonizes their voice with the author's to produce writing the author could plausibly have produced. A great editor suppresses their own voice to serve the author's. AI, trained on the same corpus for every user, cannot harmonize with any individual author. It applies the same patterns regardless of who is prompting it. Every writer using AI as a ghostwriter produces indistinguishable prose.
Readers have learned to test for authenticity because homogenized AI-written content has flooded the attention market. An author's unconventional punctuation, neologisms, and grammatical idiosyncrasies are the signals readers use to confirm a human wrote something worth their attention. AI ghostwriting removes those signals. An AI editor, by contrast, improves structure and fact-checks claims without touching the voice the author built.
Conditions
Holds when: the writer has an established voice that readers recognize and value; authenticity is a reader expectation in that medium (newsletters, essays, opinion journalism, personal blogs).
Fails when: the content is purely functional (technical documentation, legal boilerplate, transactional email) where voice is irrelevant and efficiency wins; no established author voice exists to preserve.
Evidence
Tunguz describes deliberately embedding personal stylistic markers into his writing to pass reader authenticity tests:
"In the age of slop, readers test authenticity. I lace my posts with subliminal sincerity — the ampersand, neologisms, grammatical plumes, mimicry."
He distinguishes the roles explicitly: "Great editors suppress their own voices to favor the author's. Ghost writers harmonize their own voice with the author's." Since AI cannot harmonize with individuals, it fails at the ghostwriting role and excels at the editing role.
Signals
- Readers mention specific phrases or stylistic quirks in replies, indicating they noticed the author's voice
- Newsletter open and reply rates hold stable when AI edits drafts but drop when AI rewrites them
- AI-edited drafts that preserve original idiosyncrasies feel authentic; AI-rewritten drafts draw reader comments about "sounding different"
Counter-evidence
For operators without an established personal voice or public brand, AI ghostwriting may produce acceptable output. Voice differentiation is only a moat if readers know what the author's voice sounds like. Anonymous brand-voice writing and corporate copy are unaffected by this critique. AI ghostwriting performs well for high-volume, low-brand-identity content where consistency matters more than individuality.
Cross-references
- ins_alpert-lossless-llm-rewrite (every LLM rewrite changes author meaning even without changing words)