Navigating the AI Era in Psychological Science Publishing
As artificial intelligence profoundly influences academic discourse, journal editors face the complex task of integrating these advanced tools into their publishing procedures. Across the seven academic journals of the Association for Psychological Science (APS), there are varied perspectives and approaches to this integration.
The managing editor of Observer, Hannah O. Brown, engaged with several APS journal editors, posing key questions on the subject. Their edited responses offer valuable insights. These experts include Jamie Cummins (Statistics, Transparency, and Rigor Editor, Psychological Science), Nicholas Eaton (Editor-in-Chief, Clinical Psychological Science), June Gruber (Editor-in-Chief, Current Directions in Psychological Science), Arturo Hernandez (Editor-in-Chief, Perspectives on Psychological Science), Rachael Jack (Editor-in-Chief, Advances in Psychological Science Open), Ulf-Dietrich Reips (Editor-in-Chief, Psychological Science in the Public Interest), and Felix Thoemmes (Editor-in-Chief, Advances in Methods and Practices in Psychological Science).
Regarding the transparency of AI use in manuscript preparation, editors expressed diverse opinions. Arturo Hernandez believes AI, while not entirely new, introduces automated text generation that can benefit non-native English speakers but risks producing generic or meaningless content. He argues that human judgment remains crucial for identifying novel contributions. Rachael Jack emphasizes author accountability, stating that AI tools, like any other, do not fundamentally change the need for authors to stand by their work's accuracy and integrity. She suggests focusing on responsible AI use and education. Nicholas Eaton critiques current disclosure policies as often incoherent and counterproductive, advocating for an 'artifact-centered' model where AI outputs are archived, promoting auditability over mere admission of AI involvement. June Gruber, however, stresses the importance of clear disclosure for all AI uses to better understand its impact on research. For peer review, Felix Thoemmes allows AI for summarizing but not for evaluation or writing reviews, underscoring the necessity of human judgment. June Gruber reiterates the need for reviewers to disclose AI use. Jamie Cummins highlights AI's potential in tasks like checking preregistrations but emphasizes that ultimate responsibility lies with human reviewers. Arturo Hernandez supports reviewers using LLMs for summarization and text analysis, provided human editors oversee the process to prevent over-reliance on tools that prioritize form over substance. In terms of editorial workflows, Rachael Jack finds AI useful for refining communication with authors, ensuring clarity and tone, which improves efficiency and trust. Arturo Hernandez, however, does not use AI in his workflow, relying on his extensive experience, but acknowledges its potential necessity with increasing submission volumes. Nicholas Eaton leverages AI for tasks like identifying thematic trends in past publications to aid in forming editorial boards, especially for broad-coverage journals. The discussion extends to AI's role in student learning, where Felix Thoemmes notes AI's ability to simplify coding, potentially at the cost of deeper learning, while acknowledging its inevitability. June Gruber warns against over-reliance on AI potentially diminishing critical thinking. Nicholas Eaton encourages strategic AI use for students after mastering fundamentals, viewing it as a tool for efficiency rather than a substitute for core skills. Jamie Cummins suggests students critically analyze AI outputs to understand its limitations, promoting a didactic approach over outright bans. Finally, concerning AI for translation, Jamie Cummins cautions about generative LLMs' stochastic nature and potential for mistranslation, advocating for expert human checks. Ulf-Dietrich Reips recommends human native speakers review translations, using methods like back-translation for quality assurance. Rachael Jack acknowledges AI's role in breaking language barriers but warns of risks like shifting meaning or false fluency, reinforcing author accountability for accuracy and careful oversight.
The integration of artificial intelligence into scholarly publishing represents a pivotal moment, demanding careful consideration from all stakeholders. Editors generally agree that while AI offers immense potential for efficiency, accessibility, and improving certain aspects of the publishing process, it also introduces significant challenges, particularly concerning accountability, transparency, and the preservation of human judgment. Moving forward, the scientific community must collaboratively establish robust frameworks, ethical guidelines, and educational initiatives that foster the responsible and effective use of AI, ensuring it enhances rather than compromises the integrity and quality of psychological science.