Discussion

When AI does the thinking, who gets authorship?

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#13609

A medical journal editor is asking where AI’s contribution to research leaves human authorship, as tools move beyond polishing prose to proposing research questions and hypotheses. His practical distinction is between AI assistance that researchers verify and take responsibility for, and AI doing the core intellectual work itself.

Watch Desk analysis

What happened

Professor Yoo Jin-hong, editor-in-chief of the Journal of Korean Medical Science, sets out the challenge in a recent editorial described by Korean Biomedical Review. A framework published in the journal in July divides AI use in research into language assistance, supplementary use and substantive use.

Yoo says the first two can be acceptable when researchers properly verify the work and take responsibility for it. Substantive use, where AI replaces core human judgement and creative work, sits outside current publishing ethics standards. Those standards bar AI from being listed as an author, but Yoo argues that this alone does not explain what human contribution is enough to justify authorship.

Why it matters

AI can now help select research topics, frame hypotheses and draft study designs and manuscripts. That makes authorship more than a question of whether a tool’s name can appear on a paper: journals also need to judge what people actually contributed and whether the research can be checked.

Yoo also flags a disclosure problem. If two researchers use AI to the same extent, the one who discloses it in detail could face scrutiny while the one who stays quiet may not. He says that risk is not, by itself, a reason to abandon disclosure, but it shows why disclosure rules need careful thought.

Our read

“AI cannot be an author” is a tidy rule for a much messier question. Researchers still need to stand behind their work, but editors need clearer ways to assess when a person has done enough of the intellectual work to claim authorship. The answer should not be to discount a sound discovery because AI helped make it; Yoo explicitly argues that scientific truth does not depend on whether a human or AI made the discovery.

What to watch

  • Whether journals define the human contribution required for authorship when AI performs substantive research work.
  • Whether AI-use disclosure rules can be applied without penalising researchers for being candid.
  • How editors and reviewers will check whether AI-assisted research was conducted transparently and can be verified.

Discussion spark: If AI proposes the research question and designs the study, what should a researcher have to contribute before they can claim authorship?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.