The Committee on Publication Ethics (COPE), whose standards inform the policies and practices of many philosophy journals and their publishers, has declared that “AI tools cannot be listed as an author of a paper.”

[Manipulation of Caravaggio’s “Saint Jerome Writing” by J. Weinberg]

AI tools cannot meet the requirements for authorship as they cannot take responsibility for the submitted work. As non-legal entities, they cannot assert the presence or absence of conflicts of interest nor manage copyright and license agreements.

Authors who use AI tools in the writing of a manuscript, production of images or graphical elements of the paper, or in the collection and analysis of data, must be transparent in disclosing in the Materials and Methods (or similar section) of the paper how the AI tool was used and which tool was used. Authors are fully responsible for the content of their manuscript, even those parts produced by an AI tool, and are thus liable for any breach of publication ethics.

COPE’s position matches up with that of Nature and other publications (see this previous post). (via Brian Earp)

In response to Nature’s earlier announcement, philosophers Ryan Jenkins and Patrick Lin of the Ethics + Emerging Sciences Group at California Polytechnic State University, raised some concerns about this kind of “simple policy”. In their report, “AI-Assisted Authorship: How to Assign Credit in Synthetic Scholarship“, they write:

Nature argues that crediting AI writers in the acknowledgements serves the goal of transparency. While this may be true in many cases, it could also help to hide or grossly understate the role and substantial contributions of AI writers to the paper, which is counterproductive to transparency.

Nature also argues AI writers should not be credited as authors on the grounds that they cannot be accountable for what they write. This line of argument needs to be considered more carefully. For instance, authors are  sometimes posthumously credited, even though they cannot presently be held accountable for what they said when alive, nor can they approve of a posthumous submission of a manuscript; yet it would clearly be hasty to forbid the submission or publication of posthumous works.

Thus, a more nuanced, middle-ground solution may be needed, as satisfying as a simple policy might be.

Jenkins and Lin suggest framing the matter around two questions.

The first concerns what they call “continuity”:

How substantially are the contribution of AI writers carried through to the final product? To what extent does the final product resemble the contributions of AI? What is the relative contribution from AI versus a human? The calculations are always difficult, even if the coauthors are human. Some journals routinely require statements of relative contribution to add clarity and nuance when multiple humans are sharing credit.

The second concerns what they call “creditworthiness”:

Is this the kind of product a human author would normally receive credit for? Consider whether the AI’s contributions would typically result in academic or professional credit for a human author. This analysis is similar to how we view student assistants: the greater the substance of their contribution to the final product, and the greater the extent to which this kind of product typically redounds to the credit of the author, the more important it is to credit the range of contributors, both human and artificial. 

You can read their report here.