Scientific publishing: To err is AI, to review is human

Abstract

Use of artificial intelligence (AI) tools by authors for scientific publications is becoming popular for help in achieving various tasks. Using AI assistance in scientific writing has a flip side, too, which calls for caution and raises serious concerns. Many AI tools being used do not understand the scientific concepts about which they are generating a particular text; thus, their scientific validity cannot be ensured. Fabrication of data and references, plagiarism, and intellectual property rights are other concerns. Hence, there exists a real risk of the creation of false and misleading information. The principal value of medical publications lies in their clinical utility, and the scientific community relies on them for decision-making in clinical practice. It is, therefore, imperative on part of researchers to use these tools carefully and responsibly. There should be a balance between the use of these tools and the critical evaluation of the content to ensure reliability. Rather than building the whole infrastructure based on AI, the authors should use it prudently for fine-tuning and decorative purposes. Authors must review and verify not only the content generated but also the references therein so that nothing absurd is included in the manuscript. Publishers have established policies for AI-assisted technologies used in the writing process. Their use should be limited to improving the language and clarity of the work.[1] Some insist on researchers to cite the name of any AI software platform, program used, the version, the manufacturer, and the date of its use.[2] There are two aspects of scientific publication. One is the technical information, and the other is the comprehension part that is based on critical human skills and adds to the uniqueness of the manuscript. Excessive use of AI may distort the latter due to its limited creativity. Younger authors and postgraduate students are particularly vulnerable to these vulnerabilities inherent in the use of AI and must take heed. If at all they must use it, they should review all the texts generated by using the natural intelligence gifted to them by nature, i.e. the computer placed in our heads. The purpose of this write-up is to simply highlight a double-edged sword that is here to stay. The astonishing pace of growth of AI in publishing has created a moving target as far as formulating guidelines for detecting AI-generated text creation is concerned. Till then, it is up to us to protect the legacy of our predecessors.

Ähnliche Arbeiten