On AI writing
bolkunov.com
October 20, 2026
2 min read
47/100
Summary
AI-generated writing can obscure an author’s distinctive style, reduce precision, and shift time costs from writers to readers, according to the author. The author’s bachelor’s thesis found that word-use statistics from short passages could identify authors reasonably well, while readers also recognize patterns in sentence structure and storytelling. The author argues that AI tends to average stylistic traits, making it harder for readers to infer a writer’s personality or context and to empathize with them. The author says language models must make assumptions to connect ideas, even when prompts are detailed, and may therefore choose broader or more ambiguous wording to avoid logical errors. Writers may miss that ambiguity because they supply intended meanings from their own understanding. The author also argues that drafting independently helps expose gaps and inconsistencies in a person’s thinking, whereas AI can produce a finished-seeming result before that scrutiny occurs. Ambiguous language and reduced stylistic context can make AI-generated text slower to read, the author says. The author estimates reading AI text takes at least one-third longer for them than text of the same length written without AI, and says many models’ verbosity further increases reading time. Any writing-time savings can disappear when multiple readers spend longer interpreting the output, particularly when ambiguity produces follow-up questions, revisions, and lower trust.
Key Takeaways
- The author argues that AI averages writing styles, making it harder for readers to identify an author’s personality, context, and perspective from their prose.
- The author says AI-generated prose can become less precise because language models make assumptions to join ideas and may use ambiguous wording to avoid logical mistakes.
- Drafting text can reveal weaknesses or inconsistencies in a writer’s reasoning, while AI-generated output may reduce opportunities for that self-scrutiny.
- The author estimates that AI-generated text takes them at least one-third longer to read than non-AI text of equal length, with model verbosity increasing total reading time further.
- The author says the time writers save with AI can be offset by readers’ added interpretation time, follow-up questions, collaborative revisions, and diminished trust.
What the discussion said
The thread is less interested in abstract literary theory than in the social damage caused when AI prose enters ordinary workplace communication. Commenters repeatedly describe PR descriptions, specifications, wikis, and planning documents that look polished but no longer feel authored by the colleague whose name sits above them. The objection is not merely aesthetic: readers feel the old bargain has been broken, where a writer has done the thinking before asking others to spend time reading. In the worst examples, generated planning material becomes long, vague, internally inconsistent, and costly for engineers to decode. There is broad agreement that current models are excessively wordy, hedge instead of committing, and produce recognizable verbal tics. Several readers think writing remains harder than coding or mathematics because those fields supply clearer correctness checks and reward signals. A smaller countercurrent argues this is not evidence of a permanent human moat: labs have concentrated on lucrative coding agents rather than prose, and better-trained models may narrow the gap. People do see value in constrained assistance such as spelling, critique, and perhaps code generation, provided humans still explain decisions to one another. Some also worry that AI is feeding its own habits back into human language, making people adopt its punctuation and bland framing while trying to avoid being mistaken for it.
Where opinion split
The sharp dispute is whether AI's awkward prose reveals a fundamental limit of machine intelligence or simply a neglected product and training target. One side sees authentic writing as inseparable from lived human thought and says the artifact of writing matters more than outsourcing it; the other says prose quality has received less investment because coding delivers clearer rewards and larger commercial returns.
Coverage & discussions
Hacker News · bolkunov.com
Community discussions
Community Sentiment
Positives
- AI-assisted critique and spellchecking can sharpen a person's own writing without replacing the person's responsibility for the ideas.
- Strong progress in coding and mathematical tasks shows that models can be genuinely useful where correctness is formalized and independently checkable.
- Restricting AI to drafts, code, or agent-facing material while keeping human-facing explanations human-authored offers a workable boundary for collaboration.
Concerns
- AI-written PRs and internal documents sever the link between a colleague and their stated reasoning, making readers distrust or disengage from material carrying that colleague's name.
- Generated specifications often inflate thin ideas into jargon-heavy prose, leaving teams to untangle contradictions and rebuild the actual problem statement.
- Models default to verbosity, hedging, and filler because concise prose demands precise commitments and revision rather than uninterrupted token generation.
- AI's recognizable stylistic habits are starting to contaminate human writing, encouraging bland framing and even making ordinary punctuation feel suspicious.
