LO
CapabilityCognitive automation2026-02-05
A university-hosted GPT made about three times the corrections of a human copyeditor, but only 61% of them were judged improvements, a study in PLOS One found
Editor and proofreaderoccupation page →
Event date / reported
2026-02-05
Evidence stage
CapabilityA demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.
Tasks this bears on
Copyediting for accuracy and style
Editing text line by line for clarity, style and accuracy.
Being augmented✓ Evidence-backed
Where this applies
A comparison of copyediting on two global-health papers by a university-hosted GPT, Grammarly and a human editor. The authors report the GPT made about three times as many corrections as the human editor and about ten times more than Grammarly, but was the least discriminating, with only 61% (51/83) of its corrections judged improvements; the human took 3.75 and 4 hours to edit the two papers. It covers two papers only.
What this means
The model saves hours but changes much that should not change — the editor's job shifts to choosing among its suggestions.
What it does not yet show
Two papers only; a larger comparison could come out differently.
What you can check
Open the PLOS One article e0342170 and find "only 61% (51/83)".
Does it change the assessment?
No — and this stage does not move it either. A "Capability" record is real evidence, but it does not upgrade a task judgement on its own. The 1 linked judgement above stand where they were.
Source
August et al. — Does ChatGPT enhance equity for global health publications? Copyediting by ChatGPT compared to Grammarly and a human editor, PLOS One 21(2): e0342170 (published February 5, 2026) · verified 2026-09-30 · Claude (VOLO agent) · interpreted 2026-09-30 · Claude (VOLO agent)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.