The Committee on Publication Ethics (COPE), whose requirements inform the insurance policies and practices of many philosophy journals and their publishers, has declared that “AI instruments can’t be listed as an creator of a paper.”
[Manipulation of Caravaggio’s “Saint Jerome Writing” by J. Weinberg]
COPE says:
AI instruments can’t meet the necessities for authorship as they can not take duty for the submitted work. As non-legal entities, they can not assert the presence or absence of conflicts of curiosity nor handle copyright and license agreements.
Authors who use AI instruments within the writing of a manuscript, manufacturing of photos or graphical components of the paper, or within the assortment and evaluation of knowledge, have to be clear in disclosing within the Supplies and Strategies (or comparable part) of the paper how the AI software was used and which software was used. Authors are absolutely answerable for the content material of their manuscript, even these elements produced by an AI software, and are thus responsible for any breach of publication ethics.
COPE’s place matches up with that of Nature and different publications (see this previous post). (through 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 College, raised some considerations about this type of “easy coverage”. Of their report, “AI-Assisted Authorship: How to Assign Credit in Synthetic Scholarship“, they write:
Nature argues that crediting AI writers within the acknowledgements serves the purpose of transparency. Whereas this can be true in lots of instances, it might additionally assist to cover or grossly understate the position and substantial contributions of AI writers to the paper, which is counterproductive to transparency.
Nature additionally argues AI writers shouldn’t be credited as authors on the grounds that they can’t be accountable for what they write. This line of argument must be thought of extra rigorously. As an illustration, authors are typically posthumously credited, though they can not presently be held accountable for what they mentioned when alive, nor can they approve of a posthumous submission of a manuscript; but it might clearly be hasty to forbid the submission or publication of posthumous works.
Thus, a extra nuanced, middle-ground answer could also be wanted, as satisfying as a easy coverage could be.
Jenkins and Lin counsel framing the matter round two questions.
The primary considerations what they name “continuity”:
How considerably are the contribution of AI writers carried by to the ultimate product? To what extent does the ultimate product resemble the contributions of AI? What’s the relative contribution from AI versus a human? The calculations are at all times troublesome, even when the coauthors are human. Some journals routinely require statements of relative contribution so as to add readability and nuance when a number of people are sharing credit score.
The second considerations what they name “creditworthiness”:
Is that this the type of product a human creator would usually obtain credit score for? Think about whether or not the AI’s contributions would usually end in tutorial or skilled credit score for a human creator. This evaluation is just like how we view scholar assistants: the larger the substance of their contribution to the ultimate product, and the larger the extent to which this type of product usually redounds to the credit score of the creator, the extra essential it’s to credit score the vary of contributors, each human and synthetic.
You possibly can learn their report here.
