Authors, Beware!

Researchers thinking about turning to assistance from artificial intelligence (AI) when developing a scientific manuscript should familiarise themselves with their target journal’s guidelines surrounding AI use and consider that outsourcing research and writing tasks to AI threatens essential intellectual virtues, warns Joaquín Fernández MD, PhD.

“The end or goal of academic life is flourishing through the cultivation of reason. Intellectual virtues are the means by which an academic can reach that end,” said Professor Fernández, quoting from an article in which Dr Timothy Daly, a bioethicist/philosopher of science, advocates for ‘low-tech’ academic virtue ethics.¹

“A misuse of AI tools could prevent achieving this main aim. AI tools can be used to increase efficiency, but do not forget that large language models are not authors or producers of science.”

Degrading intellectual virtues

Prof Fernández explained that as a foundation for urging academics to resist delegating tasks to AI, Dr Daly detailed the negative impacts on five key intellectual virtues—scientific knowledge, intuitive reason, practical wisdom, philosophic wisdom, and technical craft/art—described in Aristotle’s Nicomachean Ethics.

“Relying on AI to perform academic tasks could weaken these intellectual virtues that are essential for academic flourishing, causing you to lose the depth of your knowledge, your ability to judge the quality of science, your ability to connect knowledge with life and society, your sense of meaningful goals and good decisions, and your ability to create work through your own reasoning,” Prof Fernández said.

Key points for authors

Turning from a philosophical discussion to practical considerations, Prof Fernández urged authors who are drafting a manuscript for publication to check the journal’s website for guidance on acceptable uses of AI and what should be avoided because different publishers have different policies. He also cautioned against asking a chatbot to create a summary of guidelines from different publishers, noting the AI tool may not present an accurate analysis of the information.

Prof Fernández reviewed the policies from select publishers on acceptable use of AI tools for writing and image creation, what uses need to be acknowledged or disclosed, and how they should be disclosed. Reinforcing the point that there are differences between publishers, Prof Fernández presented statements on these topics from Wiley, Elsevier, IEEE, Nature Portfolio, Public Library of Science, Springer, and the British Medical Journal.

For example, when it comes to image development, Prof Fernández said Elsevier and Springer Nature typically have the most restrictive guidelines on AI use. Nevertheless, he advised caution, regardless of the targeted publication.

“Some publishers accept AI-generated images, but with exceptions,” he said. “In general, AI-generated scientific images are prohibited or strongly restricted, and so the safest approach is to avoid them unless the journal clearly allows them.”

Instructions on how to disclose AI use also differ among publishers. For example, in addition to requiring authors to name the specific AI tool or service used, Wiley requires the date of access.

Publishers also have instructions on AI use for reviewers. They, too, differ across publishers, but as a rule, Prof Fernández said AI use should be avoided during the reviewing process.

As AI continues to evolve, policies on its use for scientific manuscripts might also change. Therefore, Prof Fernández advised authors to periodically review journal guidelines to be sure they are following the most current information.

Prof Fernández spoke on this topic at the 2026 ASCRS annual meeting in Washington, DC.

Joaquín Fernández Pérez MD, PhD is CEO and Medical Director in the department of ophthalmology at Qvision, Vithas Hospital, Almería, Spain. He is the Secretary of the ESCRS. joaquinfernandezoft@qvision.es

1. Daly T. J Acad Ethics, 2026; 24: 13.