When AI writes the words, does the reader pay the time?
style
A writer's repeated way of choosing words and building sentences.
precision
How exact and clear a statement is.
ambiguity
A lack of one clear meaning.
What happened
Dima Bolkunov, a writer who published an essay about writing, examines what changes when AI produces most of a text. He defines AI writing as text with at least 50% of its words generated by AI. His argument is not simply that machines make grammar mistakes. He says AI can erase clues about the writer, blur meaning, and move work from the writer to the reader.
Background: style carries information
Bolkunov's bachelor's thesis asked whether a short paragraph could reveal its author. He compared word-use statistics with texts by other authors. The simple approach worked surprisingly well. He argues that readers notice more than vocabulary. Sentence structure and storytelling habits can also reveal a person. He says AI tends to average out these signals. A reader may see less of the writer's character and context. That can make empathy harder, even when the reader has never met the author.
Why precision and effort matter
When information is missing, AI must make assumptions to connect ideas. Even a detailed prompt cannot state every intended meaning. The system may choose broader, safer wording to avoid logical errors. The writer may miss the ambiguity because the intended meaning is already clear in the writer's mind. The reader lacks that private context and must repair the meaning while reading.
Writing also tests the writer's own ideas. Putting thoughts into sentences can expose gaps or contradictions. If AI produces a finished result immediately, the writer may lose that test. Bolkunov says he reads AI-generated text at least one-third more slowly, even when the length is equal. He blames the extra work of resolving vague words. He also says many models are highly verbose.
What is confirmed
The source confirms an argument, not a controlled study of all AI writing. Its useful contribution is the chain it proposes: less personal style can mean less context; less precision can mean more work for readers. The supplied candidate snapshot lists 76 points and 51 comments on Hacker News. Those numbers show community attention, not accuracy. Read the original essay and the Hacker News post.
What remains unknown
The essay does not show how much human editing restores style or precision. It does not compare different readers, prompts, models, or genres. It also does not measure whether extra questions, revisions, and weaker trust erase the writer's time savings. Bolkunov's one-third estimate is personal experience, not a population result.
What to watch next
The practical question is not only whether people use AI. It is who checks the meaning, who edits the final words, and who pays the reading cost. Future comparisons should measure writer time and reader time together. They should also count follow-up questions, revisions, and whether readers still trust the named author.
When AI writes for someone, who does the extra work?
📰 Full story: When AI writes the words, does the reader pay the time?
AI can save a writer time, but readers may spend more time understanding the result.
style
The special way a writer chooses words and sentences.
prompt
Instructions given to an AI tool.
💡 The gist
- Bolkunov calls text AI-written when AI creates at least half its words.
- He says AI can hide a writer's style and blur exact meaning.
- Hacker News discussed it widely. Attention does not prove the essay correct.
Dima Bolkunov wrote an essay about how AI changes writing. He defines AI writing clearly. AI must create at least 50% of the words. This definition is his choice for the essay.
He begins with his university research. He studied whether a short paragraph could reveal its author. He compared word choices across writers. The simple method worked better than expected. People also notice sentence shapes and storytelling habits. These clues help readers sense who wrote something.
Bolkunov says AI can smooth out those clues. The writing may sound less personal. Readers may lose information about the writer's background. That can make understanding the writer harder.
The essay gives another reason. AI must fill gaps when a prompt leaves something unsaid. It may guess and choose words with a wide meaning. This can make a sentence less exact. The writer may not notice. The writer already knows the intended idea. The reader does not know it. The reader must stop and repair the meaning.
Writing can also test the writer's thoughts. A person may notice gaps after putting ideas into sentences. AI can jump straight to a finished answer. That may remove a useful thinking step.
Bolkunov says he reads AI text at least one-third more slowly. He says unclear wording causes extra work. He also says many models write too much. This is his experience, not a test of every reader.
The essay's main question is about shifted effort. A writer may save time. Readers may spend more time understanding, asking questions, or requesting changes. The Hacker News post drew attention. The candidate snapshot lists 76 points and 51 comments. Those numbers show interest, not proof.
The essay does not show which edits restore a writer's style. It does not compare models, prompts, or readers. The next useful test would measure both writing time and reading time. It should also measure follow-up questions and revisions.
💬 Why can AI writing feel hard to read?
Commenters said AI writing often feels long, vague, and unlike a real person. They also said AI can help with proofreading, and that better human-like writing may still be possible.
- AI can change how people write. But using it to check spelling or critique a draft can help the writer learn, which is different from having AI write the whole thing.
- AI writing may use familiar patterns, extra words, vague language, and a voice that does not sound personal. Some commenters noted that humans used vague writing long before AI.
- If a workplace document does not sound like the colleague named on it, readers may trust it less. One reader reported that AI text took at least one-third longer to read; this was one person’s experience.
- An LLM predicts likely next words. Good editing means going back and rethinking the text, which AI may not do well. One commenter estimated that confusing AI plans caused about 100 times more human work than they saved; this was a self-report.
- Some people value writing because the person had to think through it. Others believe AI may write naturally one day if more time and resources are invested.
initial digest at 51 comments (revision 1). We fetched 41 comments and sampled 41 across the thread. These are HN users’ reports, not independently verified facts.
Can AI make writing easier but reading harder?
📰 Full story: When AI writes the words, does the reader pay the time?
AI can write words for one person, while another person does more thinking.
AI writing
Words made by an AI tool.
Hacker News
A website where people discuss technology.
Dima Bolkunov (a person who writes about words) made a small article. He calls it AI writing when AI makes half or more words.
AI can make sentences for a person. The sentences may sound smooth. They may not sound like that person. People's writing shows their word choices and sentence shapes. Those clues can help us imagine the writer. AI may hide some of those clues.
When people write, they can notice thoughts that do not fit. AI can jump straight to a finished answer. The writer may miss those problems. AI may also guess missing details. Readers then need to guess the meaning. The writer saves time. The reader may spend more time.
Hacker News, a technology discussion website, talked about the article. That shows interest. It does not prove the article is right. Bolkunov's idea is not a test of everyone. We still do not know how every reader will react.
💬 Why does AI writing feel strange?
AI can help with small writing jobs, but its words can feel too long and not very personal. People disagree about whether it will get better.
- AI can fix spelling, but its writing may sound alike, use too many words, and not sound like the coworker whose name is on it. People also wrote vague sentences before AI.
- One reader said AI text took at least one-third longer to read. That is one person’s report.
- Good writing means going back and thinking again. AI guesses the next word, so it may miss the main point. One person estimated that AI plans caused about 100 times more human work than they saved; that was a self-report.
- Some people think the thinking done while writing is the real value. Others think AI may learn to write more naturally if people invest more in it.
initial digest at 51 comments (revision 1). We fetched 41 comments and sampled 41 across the thread. These are HN users’ reports, not independently verified facts.
💬 Can AI writing replace a writer’s voice and thinking time?
Commenters argued that the problem with AI writing is not just punctuation habits, but verbosity, empty formulas, and the loss of a recognizable authorial voice. Others noted that AI is useful for critique and proofreading, and may become more natural with further investment.
initial digest at 51 comments (revision 1). We fetched 41 comments and sampled 41 across the thread. These are HN users’ reports, not independently verified facts.