When AI helps you draft a comment, how much of it is still your voice, and how often do you keep its version without changing much?

This explores how much of an AI-assisted comment still belongs to the person who posts it: their wording, their stance, and how they come across. No study in the collection counts words directly. It does measure something more revealing, which is how little of the writer survives once AI is involved. Writers edited AI-generated paragraphs only 23% of the time, and even the edited versions stayed 96% similar to what the AI wrote Do writers actually edit AI-generated text before publishing?. In practice, the AI's draft is usually what gets published.

You might expect that to be fine if writers only accept drafts that say what they meant. But in a study of 4,503 cases, writers chose the AI version of their own paragraph 63% of the time, and 52% said it reflected their views better than what they had written themselves. That happened even though the AI versions consistently shifted the original stance Do writers actually prefer AI-edited versions of their own text?. So the writer's approval can't settle whether the words are still theirs, because the drift happens without them noticing. What does change their sense of ownership is control. People feel more ownership when they actively steer the generated text. A model personalized to their style doesn't raise that feeling Does user control over AI text shape feelings of ownership?.

Readers, however, pick up on what's been lost. Across 2,939 writers and 11,091 readers, AI assistance shifted all 29 measured traits of how the writer came across. Writers seemed more confident, more extreme, more agreeable, and more privileged Does AI writing assistance change how readers perceive the writer?. They were read as far more educated, wealthier, and more likely to be native English speakers than they were, which the researchers call 'identity laundering' Does AI writing make authors seem more privileged than they are?. The pull runs in one direction. Indian writers using GPT-4o autocomplete drifted toward Western phrasing and cultural references, while American writers got bigger productivity gains Do AI writing assistants push non-Western writers toward Western styles?. The further a writer starts from the model's default voice, the more of that voice gets replaced.

In comment sections specifically, a 680-person experiment found that AI commenting tools produced longer comments and more participation. Readers still judged the comments generic and less authentic, and that impression spread to conversations among people who hadn't used the tools Do AI writing tools improve online discussion or degrade it?. Engagement metrics don't capture any of this. Machine-generated Reddit comments, with their recognizable assistant-style warmth, draw about as much engagement as human ones Does machine-generated text get penalized in online engagement?. Spotting the difference doesn't work well either. Across 25 million Hacker News and Reddit comments, the features that actually separate AI prose from human prose didn't predict which comments got called 'slop' Do AI slop accusations actually detect AI text?. One paper claims heavily rewritten messages also slip past AI detectors, but it shows only that writing styles converge and never tests a detector Do rewrites that hide authorship also fool AI detectors?.

The short answer is less than writers think, and the part that goes missing is the part that marks them as individuals. Wording mostly passes through unedited. Stance and persona drift toward one confident, polished, privileged voice. The writer's own sense of authorship is the least reliable measure of all.