AI writing tools are aplenty. And the usage of AI is now fairly prevalent in the academic landscape. While AI improves wording and helps identify surface-level inconsistencies, it is not reliable for verifying the technical accuracy, ethical compliance, scientific soundness, and how compelling a manuscript is for a journal\u2019s readership. <\/p>\n\n\n\n
Therefore, research manuscripts deserve crucial human-driven checks before the final journal submission<\/strong><\/a>. <\/p>\n\n\n\n 1. Verify the Research Question\u00a0<\/a><\/p>\n\n\n\n 2. Check Journal Fit\u00a0<\/a><\/p>\n\n\n\n 3. Ensure Methodology is Reproducible\u00a0<\/a><\/p>\n\n\n\n 4. Evaluate the Statistics and Interpretation\u00a0<\/a><\/p>\n\n\n\n 5. Check Citation Accuracy\u00a0\u00a0<\/a><\/p>\n\n\n\n 6. Validate the Strength of Evidence<\/a>\u00a0\u00a0<\/p>\n\n\n\n 7. Review the Discussion Section for Balance\u00a0<\/a><\/p>\n\n\n\n 8. Ethics, Consent, and Participant Anonymity\u00a0<\/a><\/p>\n\n\n\n 9. Authorship, Contributions, Disclosures\u00a0<\/a><\/p>\n\n\n\n 10. Originality, AI use, submission compliance\u00a0<\/a><\/p>\n\n\n\n The ability of AI to sound confident can mask the underlying problems in writing. For example, AI writing can be so coherent that authors may fail to recognize that the research question is too broad or does not align with the study. So, check whether the central claim of the study is sufficiently meaningful. <\/p>\n\n\n\n The research question should be clearly stated, with appropriate alignment between the study objectives, hypotheses, methods, findings, and conclusions. Verify that the evidence supports your claims and your manuscript must answer why your research matters. <\/p>\n\n\n\n When checking for language, AI tools do not verify the suitability of your manuscript to a journal. Of course, there are other tools and platforms that can help shortlist journals. But do a manual review of the journal\u2019s aims and scope, recently published articles, preferred article type, word limit requirements, target readership, and accepted subject areas. This will give you clarity on how well your manuscript fits the journal. <\/p>\n\n\n\n No one understands the context of your research better than you. But your research methodology is meant to be reproduced and repeated by other qualified researchers. AI checks could miss these, but you must validate the accuracy of <\/p>\n\n\n\n Studies involving statistics and interpretation require careful assessment of whether the correct statistical tests were implemented for the data and study design. Sometimes, a result may be statistically significant, but is it scientifically important? That\u2019s something to be determined by a human checker rather than an AI tool. <\/p>\n\n\n\n Also, AI tools cannot evaluate whether missing data or negative findings have been reported. If, during analysis, you come across contradictory results or null findings, you as authors should take a call on how best to report them tactically without weakening your research argument. Such checks are crucial and warrant expert human intervention. A statistician or experienced subject expert may identify problems that a general-purpose AI tool fails to detect. <\/p>\n\n\n\n AI tools are notorious for fabricating citations [1]. So it is mandatory for you to validate the accuracy of citations listed in your work. However, what matters is more than simply ensuring that the citation is real. <\/p>\n\n\n\n You want the reference cited to truly support the claim you are making. This means a deep dive into your cited work to ensure that it strongly supports the statement. Also check whether the latest, recent studies have been cited. Older works with outdated research methodology may no longer be entirely relevant. <\/p>\n\n\n\n Another thing to watchout for: accidentally citing retracted papers [2]. Keep yourself updated on whether there have been retractions of papers in your field. Have you cited those in your works before? Do they need a relook? Authors often tend to build their research on their own previous works. And if you have cited retracted papers in your previously published papers, you do not want to carry them forward unintentionally. <\/p>\n\n\n\n Another drawback of AI tools is they cannot correctly assess if your claims align with the evidence presented. For example, in trials involving drug interventions, over-stating the effect of new drugs on helping cure a medical condition will be deemed unethical. You are expected to report what you observed: the actual extent to which a condition can be cured and avoid exaggerating the impact to show that the newly introduced drug is exceptional. <\/p>\n\n\n\n The Discussion section is where the research comes together. It is not about what your investigation found but about what those findings mean. <\/p>\n\n\n\n Maintain proper balance between the implications and limitations of the study. AI tools might recommend underplaying the limitations to avoid showing that your study lacked something. Contrarily, mentioning the limitations adds credibility to your study, letting reviewers and readers know that there\u2019s more to explore in the future. <\/p>\n\n\n\n At the same time, do not overstate the study implications. For instance, your study on improving driving safety in automated vehicles does not guarantee complete avoidance of road accidents! Factors like road conditions, driving skills, physical infrastructure, operational constraints, environmental conditions ,all come in to play. Therefore, the implication should be along the lines of creating better driver awareness through advanced driving safety systems rather than avoidance of accidents. <\/p>\n\n\n\n Top publishers clearly state in their AI use guidelines that an AI tool cannot be held responsible for unethical research practices. As authors of the paper, the responsibility of conducting and reporting research ethically lies with you. <\/p>\n\n\n\n Check whether <\/p>\n\n\n\n No journal recognizes AI as an author! Therefore, do not say you used AI to author your manuscript. However, appropriate AI disclosures<\/a> must be made to clarify in what capacity AI contributed to preparing the manuscript. <\/p>\n\n\n\n Also, check whether all other author contributions and disclosure statements are included as required. Things like order of author names, accurate affiliations, contribution categories, funding sources, conflicts of interest\u2014everything must be checked before journal submission. <\/p>\n\n\n\n Finally, perform a thorough integrity and formatting check of your manuscript. Confirm that your manuscript is not under consideration elsewhere. Issues like duplicate publication or significant overlap with previously published papers are problematic. <\/p>\n\n\n\n Also check for overall submission compliance. Have you included the necessary supplementary files? Is the cover letter complete and accurate? Does the submitted file format adhere to the journal\u2019s requirements? Is your paper anonymized? Is the word count specified? These checks vary journal to journal. So keep a customized list and tick off the items as you complete the checks. <\/p>\n\n\n\n Again, prioritize checking for proper disclosures. Especially when it comes to AI disclosure statements. Journal policies of AI use are subjective. Which means the requirements of one journal may not be the same as those of another. So ensure you have read the journal\u2019s policies for AI use and followed the guidelines suitably. <\/p>\n\n\n\n References<\/strong> <\/p>\n\n\n\n 1. <\/strong>Fabricated citations: an audit across 2.5 million biomedical papers <\/p>\n\n\n\n1. Verify the Research Question <\/strong> <\/h2>\n\n\n\n
2. Check Journal Fit<\/strong> <\/h2>\n\n\n\n
3. Ensure Methodology is Reproducible<\/strong> <\/h2>\n\n\n\n
\n
\n
\n
\n
\n
\n
4. Evaluate the Statistics and Interpretation<\/strong> <\/h2>\n\n\n\n
5. Check Citation Accuracy<\/strong> <\/h2>\n\n\n\n
6. Validate the Strength of Evidence<\/strong> <\/h2>\n\n\n\n
7. Review the Discussion Section for Balance<\/strong> <\/h2>\n\n\n\n
8. Ethics, Consent, and Participant Anonymity<\/strong> <\/h2>\n\n\n\n
\n
\n
\n
\n
\n
9. Authorship, Contributions, Disclosures<\/strong> <\/h2>\n\n\n\n
10. Originality, AI use, Submission Compliance<\/strong> <\/h2>\n\n\n\n