Artificial intelligence tools such as ChatGPT are increasingly becoming part of academic research and scholarly writing. Researchers, students, faculty members, and professionals use AI to generate ideas, improve language, summarise information, organise arguments, and refine manuscripts. These tools can save time and improve productivity, but they also introduce important questions related to originality, authorship, transparency, accuracy, plagiarism, confidentiality, and academic integrity.
The key issue is not simply whether researchers should use ChatGPT. The more important question is how researchers can use ChatGPT ethically and responsibly without compromising the credibility of their work.
Used carefully, ChatGPT can support the research process. Used irresponsibly, it can introduce fabricated references, inaccurate claims, undisclosed authorship assistance, inappropriate paraphrasing, and even research misconduct. Researchers therefore need clear principles for responsible AI-assisted writing.
Use ChatGPT as an Assistant, Not as the Researcher
The most important principle is to treat ChatGPT as a support tool rather than as a substitute for scholarly judgement.
Research involves identifying a meaningful problem, reviewing evidence, designing an appropriate methodology, collecting or analysing data, interpreting results, and developing conclusions. These activities require intellectual responsibility and subject expertise.
ChatGPT can help researchers organise information or improve presentation, but it should not independently determine what a study means.
For example, a researcher can ask ChatGPT to suggest alternative ways of structuring an introduction. However, the researcher should decide which arguments are scientifically appropriate and should verify every factual statement included in the final paper.
The human researcher must remain responsible for the manuscript.
Use AI for Brainstorming and Idea Development
One ethical use of ChatGPT is brainstorming.
Researchers may use it to explore potential research questions, identify broad themes, generate keywords, organise conceptual categories, or consider alternative ways of explaining a problem.
Suppose a researcher is studying public transport accessibility. ChatGPT may help generate possible themes such as affordability, first- and last-mile connectivity, reliability, safety, comfort, or service frequency.
These suggestions can help researchers think more broadly.
However, AI-generated ideas should not automatically be treated as research gaps. A genuine research gap must be established by reviewing published literature and understanding the state of knowledge in the field.
AI can generate possibilities, but the researcher must verify them.
Improve Language Without Changing Scientific Meaning
ChatGPT can be useful for language editing, particularly for researchers writing in a second language.
It may help improve:
grammar,
sentence structure,
academic tone,
readability,
paragraph transitions,
conciseness,
clarity,
and consistency.
For example, a researcher may write a technically correct paragraph that is difficult to read. ChatGPT can help restructure the paragraph while preserving the original meaning.
However, researchers should carefully compare the revised text with the original.
AI may occasionally change technical meaning, exaggerate conclusions, simplify important qualifications, or introduce unsupported statements.
Language improvement should therefore always be followed by human verification.
Never Trust AI-Generated References Without Verification
One of the most serious risks of using ChatGPT in academic writing is fabricated or incorrect references.
AI systems may sometimes generate references that appear authentic but do not actually exist. They may also produce incorrect author names, article titles, journal names, publication years, page numbers, or DOIs.
Researchers should never insert a reference into a manuscript merely because ChatGPT provided it.
Every citation should be verified through reliable academic sources such as:
Scopus,
Web of Science,
PubMed,
Crossref,
Google Scholar,
publisher websites,
or recognised academic databases.
Researchers should ideally retrieve and read the original source before citing it.
A reference is not trustworthy simply because it looks academically formatted.
Do Not Use ChatGPT to Fabricate Data
Artificial intelligence should never be used to invent research participants, survey responses, experimental observations, interview transcripts, statistical results, laboratory findings, or field data.
Fabricated data represent a serious violation of research integrity.
If a researcher has collected responses from 300 participants, the analysis should be based on those actual responses.
ChatGPT may assist in explaining how statistical techniques work or help interpret genuine analytical outputs, but it should not create fake findings and present them as empirical evidence.
Similarly, simulated data should only be used when simulation is an explicit and legitimate part of the research methodology, and this must be clearly disclosed.
Transparency is essential.
Be Careful When Using ChatGPT for Literature Reviews
ChatGPT can help researchers understand concepts, generate search terms, classify themes, or summarise text that researchers provide.
However, it should not replace a systematic search of academic databases.
A literature review must be based on identifiable and verifiable sources.
Researchers conducting systematic reviews should maintain records of:
databases searched,
search strings,
publication dates,
inclusion criteria,
exclusion criteria,
screening stages,
and final included studies.
ChatGPT may support some parts of this workflow, but the evidence base should remain transparent and reproducible.
Researchers should avoid writing statements such as "most studies show..." unless they have actually reviewed the relevant literature.
Use ChatGPT Carefully for Paraphrasing
Paraphrasing is another area where researchers should exercise caution.
ChatGPT can rewrite sentences, but rewriting someone else's ideas does not remove the obligation to cite the original source.
Suppose a researcher takes a paragraph from a published article and asks ChatGPT to rewrite it completely. Even if the resulting wording is different, the underlying idea still belongs to the original authors.
Proper citation remains necessary.
Ethical paraphrasing involves understanding the original material, expressing the concept accurately in new language, and acknowledging the source.
AI should not be used to disguise copied material or manipulate similarity-detection systems.
Verify Statistical and Methodological Advice
Researchers increasingly ask ChatGPT questions about statistical analysis, research methodology, SPSS, R, Python, SmartPLS, regression, structural equation modelling, qualitative coding, and other analytical techniques.
This can be useful for learning.
However, statistical advice from AI should not be accepted without verification.
ChatGPT may provide an incorrect formula, recommend an inappropriate test, misinterpret a coefficient, or overlook important assumptions.
Researchers should verify analytical decisions using textbooks, methodological literature, software documentation, experienced statisticians, or qualified supervisors.
AI can explain methods, but methodological responsibility remains with the researcher.
Protect Confidential and Sensitive Information
Researchers should also consider data privacy when using generative AI.
Unpublished manuscripts, confidential reviewer reports, participant information, personal identifiers, proprietary datasets, institutional documents, or commercially sensitive information should not be uploaded casually to external AI platforms.
Before sharing research material with an AI system, researchers should consider:
institutional data policies,
research ethics approval,
informed consent,
confidentiality requirements,
copyright restrictions,
and data-protection obligations.
Removing names does not always guarantee anonymity, especially when datasets contain multiple identifying characteristics.
Researchers working with sensitive material should use approved systems and follow institutional guidance.
Disclose AI Use When Required
Transparency is increasingly important in scholarly publishing.
Many journals, universities, and publishers have established policies concerning generative AI.
Researchers may be required to disclose whether AI was used for language editing, manuscript preparation, data analysis, coding, image creation, or other research-related activities.
The exact disclosure requirements vary.
A researcher might write a statement such as:
"Generative AI was used for language editing and improving readability. All scientific content, interpretations, references, and conclusions were reviewed and verified by the authors."
Researchers should check the specific policy of the target journal before submission.
Disclosure is especially important when AI has materially influenced the research process.
Do Not List ChatGPT as an Author
Authorship carries responsibility.
An academic author is expected to take responsibility for the accuracy, originality, ethics, and integrity of the work.
An AI system cannot independently accept responsibility, approve a final manuscript, disclose conflicts of interest, respond to ethical allegations, or take accountability for errors.
For this reason, ChatGPT should be treated as a tool rather than as a co-author.
Human researchers remain accountable for everything submitted under their names.
Keep Records of Your Research Process
Maintaining records can strengthen research transparency.
Researchers should retain:
original drafts,
research notes,
source files,
datasets,
statistical outputs,
literature-search records,
reference libraries,
analysis scripts,
and important AI prompts where relevant.
This documentation can help demonstrate how the research developed.
It can also be useful when responding to reviewers, supervisors, editors, or research-integrity questions.
Good documentation is especially valuable as AI becomes more integrated into academic workflows.
Use AI to Improve, Not Replace, Critical Thinking
Perhaps the greatest danger of excessive dependence on AI is not plagiarism but reduced critical thinking.
Researchers may become tempted to accept convenient explanations instead of examining evidence.
Good research requires questioning assumptions, comparing competing explanations, identifying limitations, recognising uncertainty, and developing arguments based on evidence.
ChatGPT can support these activities by generating alternative perspectives or asking critical questions.
However, the researcher should evaluate those responses rather than accepting them automatically.
The goal should be AI-assisted scholarship, not AI-dependent scholarship.
A Simple Ethical Checklist for Researchers
Before using AI-generated material in a research paper, researchers should ask:
Is this statement accurate?
Can I verify it from a reliable source?
Have I cited the original source where necessary?
Did the AI change the meaning of my argument?
Does my university or journal allow this type of AI use?
Do I need to disclose that AI was used?
Have I protected confidential data?
Am I personally willing to take responsibility for this content?
If the answer to any of these questions is uncertain, the material should be reviewed before submission.
Conclusion
ChatGPT can be a valuable tool for research paper writing when it is used responsibly. It can assist with brainstorming, language improvement, organisation, explanation, summarisation, and technical support. It may also help researchers communicate complex ideas more clearly and work more efficiently.
However, ethical research requires more than efficient writing.
Researchers must verify information, protect confidential data, avoid fabricated references and data, cite sources properly, disclose meaningful AI use where required, and remain fully responsible for the final manuscript.
The safest principle is simple: use ChatGPT to support your research, but never allow it to replace evidence, scholarly judgement, transparency, or academic responsibility.
For research-support organisations such as EduPub, promoting responsible AI use can help researchers benefit from emerging technologies while protecting the standards of originality, credibility, and integrity that scholarly publishing depends upon.
