AI evidence extraction

Human‐ and AI‐based authorship: Principles and ethics

Jaime A. Teixeira da Silva, Panagiotis Tsigaris · Learned Publishing · 2023

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

1/4

Quality (LMQS)

Evidence

Citations

1.14

FWCI

Top 10%

Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1002/leap.1547

Methodology & findings

Study design

Narrative review combined with direct engagement with ChatGPT through structured queries.

Research paradigm

Critical interpretivism with philosophical inquiry into authorship ethics

Author conclusions

The authors conclude that "Do the ICMJE, COPE and CSE guidelines and recommendations related to authorship need to be revisited and corrected to accommodate the functions, rights and representation of AI? We believe that such guidelines need to be adjusted, if necessary, provided that this is the will of the community to adapt and adopt new values. That debate is now taking place, and it will only be a matter of time before new or adjusted guidelines appear." They also conclude: "Currently, ChatGPT does not want to be an author or co-author" and emphasize that institutions "still need to offer more guidance regarding the issue of AI authorship and its role within the debate of honorary and ghost authorship."

Risk of bias

Author subjectivity in evaluating whether AI can satisfy CRediT criteria; Temporal limitation: assessment based on ChatGPT free version as of February-April 2023, with rapid model improvements potentially occurring afterward; Selection bias: focus primarily on ChatGPT rather than other AI systems; Confirmation bias potential: authors appear predisposed against AI authorship based on philosophical grounds; Author bias in subjective assessment of ChatGPT capabilities against CRediT criteria; Limited scope: interactions with single AI system (ChatGPT) at single point in time; Potential confirmation bias in selecting questions posed to ChatGPT; Interpretive bias in translating ChatGPT's responses; Funding bias not disclosed, though Bronze OA status suggests potential institutional or publisher influence

Open questions raised

The authors identify several future research directions: (1) whether predatory journals pay attention to authorship validity (described as "a topic that has yet to be explored"); (2) the need for updated ICMJE, COPE, and CSE guidelines to accommodate AI authorship; (3) the role AI identities should play in authorship debates; and (4) the ethical implications when humans treat AI as ghost authors.
The role of AI authorship in predatory journals and publishers, described as "a topic that has yet to be explored"
Need for clearer guidance from institutions like ICMJE regarding AI authorship and its relationship to honorary and ghost authorship
Unclear moral and legal authority to make judgments about AI authorship rights
Need for assessment of AI role in papers with dozens, hundreds, or thousands of human authors
Distinction between technical and intellectual contributions in authorship frameworks

Extracted from: pdfAgreement 66%

Explore related topics

AI Manuscript Writing Academic Writing Conceptual