Dear reviewers and authors of Oncoradiology
Firstly, we sincerely thank you for your continuous support and dedication to Oncoradiology. Your professional knowledge and hard work have played a crucial role in enhancing the academic quality and influence of the magazine.
Recently, we have noticed that some experts have started to us generated artificial intelligence (AI) models for review work, which has caught our attention. The editorial department also tested the AI models for reviewing some papers and invited experts to evaluate the AI review comments. We found that although there is some potential, there are still certain deficiencies and risks in the direct review comments provided by the AI models. Therefore, Oncoradiology forbids using the AI models for peer reviewing of papers and for writing as well, mainly for the following reasons:
1. AI models lack the ability to evaluate research innovation: The evaluation of innovation largely relies on the depth and breadth of experts' professional knowledge. Currently, AI models lack a deep understanding of the research field and unique insights into research problems, as well as training in the latest knowledge in the field, resulting in inaccurate innovation evaluation.
2. The AI models tend to be positive in research review: The AI models perform outstandingly in data extraction and summarization, so the review of papers mainly relies on summarizing and extracting the content of the article itself. The author's writing of the article tends to be positive, so it is easy to have a biased positive evaluation and rarely returns the manuscript. As for the authors who use AI to create an articles, who might leads to plagiarism.
3. The AI models normally have a relatively one-sided approach to research modification suggestions: When proposing research modification suggestions, the AI models mainly relies on the limitations mentioned by the author in the discussion, and does not combine its own knowledge and existing progress to propose unique modification suggestions.
4. Convergence and Qualitative Review Comments on General AI Models: The review comments on AI models lack individuality and depth, have a single mode and formatted form, but the review quality is low and cannot provide targeted and constructive feedback to authors.
5. Other issues: There are also other hidden dangers in composing by AI models, such as content leakage, data security, review bias, review illusions, and unclear responsibilities.
Peer review is the core of academic publishing and a key link in ensuring the innovation, scientificity, and ethics of papers. Based on their academic accumulation and practical experience, the reviewing experts comprehensively evaluate the dimensions of research design, data reliability, and logical conclusions, and provide constructive suggestions. This process is not only a test of the author's academic achievements, but also an important way for the academic community to self-correct and make progress together. The independent thinking and professional judgment of experts, and articles created by authors are irreplaceable and the cornerstone of the continuous development of Oncoradiology. At present, AI models are unable to undertake the role of expert peer review and authorship.
At the same time, we suggest that peer reviewers should maintain a cautious attitude when using artificial intelligence models to assist in peer review, and inform the editorial department. At the same time, experts should be responsible for reviewing the opinions of the assisting reviewers, fully utilizing their professional judgment and unique insights to ensure the fairness and accuracy of the review results.
Thank you again for your support and contribution to Oncoradiology!
Editorial Office of Oncoradiology
13th Feb. 2025
