Scholarly publishing continues to evolve as researchers, publishers, editors, and peer reviewers navigate the growing role of artificial intelligence and changing expectations around peer review. Recent discussions have focused on a common question: how can the publishing community adopt new approaches without losing the accountability, transparency, and human judgment that underpin scholarly communication? Here are four recent developments shaping that conversation.
Robert Braun’s article in Nature examines responsibility for AI-assisted science
In Nature, Robert Braun explores what happens to authorship and accountability when generative AI becomes part of the research and writing process. AI tools are increasingly being used to search and summarize literature, generate ideas, draft responses to reviewers, and improve written text. The article points out that while these activities can contribute significantly to scholarly work, AI cannot explain, defend, or take responsibility for its output in the way a human contributor can.
This raises questions about whether existing authorship frameworks are equipped to capture AI-mediated work. The article examines CRediT, the Contributor Roles Taxonomy, noting that while it provides a more detailed picture of human contributions, it was not designed to account for AI involvement. The article suggests that the arrival of AI provides an opportunity to reconsider how scholarly contributions are recognized, including whether work traditionally assigned to research assistants and other contributors receives appropriate credit. Read the full article here
Wenyu Li and colleagues study how LLMs perform as peer reviewers
A study published in New Biotechnology by Wenyu Li and colleagues examines whether large language models can provide reliable peer review in biotechnology. The researchers evaluated GPT-5, Qwen-Plus, and Gemini 2.5 Pro using 763 preprints, including 398 with open peer reviews, as well as 12 grant proposals. The AI systems produced substantive and structured feedback, with particular attention to experimental design and statistical analysis.
However, the study found notable differences between AI-generated and human reviews. AI reviewers tended to be more lenient, were less likely to question how papers were positioned within the field or request additional citations, and generally rated grant proposals more favorably than human reviewers.
The researchers also examined AI detection tools and found that they were not reliable at identifying AI-generated review comments, particularly when the text was lightly reworded. The study highlights privacy and copyright concerns associated with using public AI systems for peer review and emphasizes the need for evidence verification, transparency, clear guidelines, and continued human oversight. Read the full article here
Helen Kara and Jenni Guthrie explore a more human approach to peer review
Writing for the LSE Impact Blog, Helen Kara and Jenni Guthrie examine whether peer review can become more relational and collaborative. The authors draw on two experiences. The Journal of Creative Research Methods explored whether authors would continue submitting if the journal adopted an open peer review model. Of 128 initial responses, 120 said they would. Feedback in favor of the approach highlighted transparency, collegiality, learning, support for less experienced researchers, respect, fairness, and collaboration.
The second example comes from the authors’ work on the Palgrave Handbook of Neuroinclusive Social Work, where editors worked closely with contributors through open feedback and dialogue. This approach allowed editors and authors to challenge ideas, clarify misunderstandings, and learn from one another while developing the work.
The authors acknowledge that relational peer review can bring challenges, including potential bias, additional time requirements, and difficulties finding reviewers. They argue, however, that these experiences demonstrate the potential for approaches that move beyond a standardized model. The article ultimately presents open and relational peer review as one possible approach for communities where dialogue, accountability, and collaboration can strengthen scholarly work. Read the full article here
COPE introduces new frameworks for publication ethics
In an interview with The Scholarly Kitchen, COPE Executive Officer Natalie Ridgway discusses the organization’s new Core Principles and Code of Conduct, which replace the previous Core Practices framework. The Core Principles are intended to provide a shared ethical foundation for the scholarly community, particularly in situations where specific guidance may not yet exist. The Code of Conduct, meanwhile, sets clearer and more direct expectations for COPE members and provides a stronger framework for accountability.
The interview addresses how journals should handle suspected misconduct. COPE distinguishes between the editor’s responsibility to assess the reliability of published research and an institution’s responsibility to determine questions of author intent or culpability. It also emphasizes transparency in post-publication notices while recognizing the importance of confidentiality and applicable legal requirements. The updated frameworks are part of COPE’s broader focus on strengthening integrity and accountability in scholarly publishing. While COPE does not describe itself as a regulator, the organization expects the revised approach to bring greater clarity around membership expectations and how concerns are handled. Read the full article here
If you’ve come across a piece lately that sparked reflection or raised important questions, feel free to share it with the ReviewerOne community.
