When AI helps you write, your drafts get better today, but does your own writing skill grow with it?
Can collaboration with GenAI preserve long-term skill development in writing work?
This explores whether writers who work alongside generative AI can keep building their own writing ability over time, or whether the help raises today's output while their underlying skill stalls.
This explores whether working with AI on writing can help writers build lasting skill, not just better drafts today. The short answer from the corpus: there's no direct long-term study of writers yet, and the closest evidence points the wrong way. When workers used generative AI on content tasks they performed much better, but when they later did similar tasks on their own, they showed no improvement. The gain stayed with the tool and didn't carry over to the person (Does AI assistance help workers learn lasting skills?). Better output while assisted is not proof that anyone learned anything.
One likely reason is how people actually use the suggestions. In one large study, writers edited AI-generated paragraphs only 23% of the time, and even their edits stayed about 96% similar to the original (Do writers actually edit AI-generated text before publishing?). Skill grows through revising, rejecting and rewriting, and that work is mostly being skipped. The skipped work also has a cost you can see on the page. AI assistance shifted how readers perceived writers on all 29 traits measured, making them seem more confident, more extreme and more privileged than they were (Does AI writing assistance change how readers perceive the writer?). Autocomplete pulled Indian writers toward Western phrasing and cultural references (Do AI writing assistants push non-Western writers toward Western styles?). So what's at risk isn't only technique. It's voice, which is arguably the hardest part of writing to develop.
The more hopeful material is about stance, meaning how a writer relates to the AI. A survey of 403 writers found that those who both collaborate with AI and treat it as a rival to outdo reported the best crafting and productivity outcomes (Does balancing rivalry and collaboration with GenAI boost writer productivity?). Note the limits: this is self-reported and measured at one point in time, so it shows a link, not a cause. It also measures perceived productivity, not skill growth. The same authors suggest adding small frictions to the interface to keep that competitive edge alive, but they never tested the idea (Can micro-frictions boost rivalry without harming collaboration?). Treat it as a hypothesis worth watching.
Some design work hints at what skill-preserving collaboration might look like: the writer stays in charge of the process. Writers who set up an AI "thought partner" in advance, deciding its role and how often it should chime in, used it to generate ideas and to monitor their own writing rather than to produce text for them (Can writers benefit from configuring AI writing partners in advance?). In shared editors, writers wanted to see each other's prompts, partly so they could check AI-generated text instead of trusting it blindly (Do writers want to see each other's AI prompts in shared editors?). Both point in the same direction: the AI as a sparring partner or a planning aid, not a ghostwriter.
The thing you might not have expected: the damage may not stay with the individual writer. In a 680-person experiment, AI commenting tools increased participation, but readers rated whole conversations as more generic and less authentic. That included threads among people who hadn't used the tools (Do AI writing tools improve online discussion or degrade it?). If many writers lean on the same assistant, the shared standard for what good writing sounds like may drift too. To be direct: the corpus can't yet tell you whether any collaboration style preserves skill over months or years. It can tell you which habits make that unlikely (accepting text unedited, letting the tool set the voice) and which stances look most promising.
Sources 9 notes
Wu et al. found that workers using generative AI performed substantially better on content tasks, but when performing similar tasks independently afterward, their performance showed no improvement. The capability did not transfer across contexts.
Writers edited AI-generated paragraphs only 23% of the time, with edits averaging 96% similarity to the original. This means AI's opinionated and distorted voice propagates with minimal human filtering before publication.
A study of 2,939 writers and 11,091 readers found AI assistance shifted every tested dimension—29 total—toward extremism, confidence, quality, agreeableness, and perceived privilege. Distortions were statistically significant and directional, not random noise.
A 118-person controlled experiment found that GPT-4o autocomplete pulled Indian essays toward Western phrasing and cultural references while delivering larger productivity gains to American participants, suggesting cultural distance from the model's training data creates unequal service and homogenizing pressure.
A survey of 403 writers found that those scoring high on both rivalry and collaboration toward GenAI reported the strongest crafting and productivity outcomes. The cross-sectional self-report design shows association, not causation, and productivity measures reflect writers' perceptions rather than objective performance.
Show all 9 sources
A survey of 403 writers proposed introducing micro-frictions to increase rivalry while maintaining collaboration, but conducted no intervention, comparison, or behavioral test. The hypothesis lacks evidence and requires longitudinal or experimental validation.
In a one-week study with 16 writers, participants successfully set up proactive AI partners by pre-configuring their roles and proactivity levels, then used the AI suggestions to generate ideas and monitor their own writing.
Sixteen paired writers showed strong preference for higher levels of prompt visibility in shared editors, valuing awareness of when, how, and where AI was used. Benefits included understanding collaborators' thinking and verifying AI-generated text, though some found full sharing intrusive and self-conscious.
In a 680-participant experiment, AI-assisted commenting tools produced longer comments and higher participation rates, yet readers perceived the content as generic and less authentic. The perceived decline in quality extended even to conversations among users who did not use the AI tools.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Understanding Reader Perception Shifts upon Disclosure of AI Authorship
- "It was 80% me, 20% AI": Seeking Authenticity in Co-Writing with Large Language Models
- Measuring and Mitigating Persona Distortions from AI Writing Assistance
- What Influences Readers' and Writers' Perceived Necessity of AI Disclosure?
- Penalizing Transparency? How AI Disclosure and Author Demographics Shape Human and AI Judgments About Writing
- Evidence-centered Assessment for Writing with Generative AI
- Show Me Your Prompts! How Writers Feel About Sharing Prompts in Collaborative Text Editors
- AI Suggestions Homogenize Writing Toward Western Styles and Diminish Cultural Nuances