Suggest, Guide, Rewrite: Comparing Forms of AI Writing Assistance

Under review. Fifty-four professional writers tried guidance, local suggestions, and complete rewrites. Guidance yielded lower human-rated quality and greater reported effort; 38 writers chose local suggestions to keep. Ownership and control comparisons were inconclusive, so preference does not establish that a tool preserves voice.

Authors: Iman YeckehZaare · Year: 2026

In a within-participant study, 54 professional writers each revised three standardized documents, drawn from two matched sets, using each assistance form once. Guide explained revision techniques without replacement prose; Suggest offered local replacements; Rewrite supplied a complete revision. Prepared responses shared an upstream analysis, but differed in content, replacement boundaries, and the work left to writers. Guide yielded lower human-rated quality and greater reported effort than either alternative. The Suggest–Rewrite quality difference and every ownership or control comparison were inconclusive, not evidence that the forms were equivalent. Thirty-eight writers chose Suggest to keep, eleven Guide, and five Rewrite. The results motivate offering local replacements for short revision tasks while assessing quality, experience, and preference separately. They neither isolate an automation effect nor establish voice preservation.

Public artifact boundary: this route exposes status, authorship, visual summary, citation metadata, and the contribution boundary; manuscript files are posted only when review and prepublication rules allow it.

The rendered manuscript page adds status, visual summary, review boundary, citation metadata, and contribution notes. Key links: home, systems, papers, manuscripts, Google Scholar, GitHub, LinkedIn, ORCID, MIT profile, and CV PDF.