JOHOR BAHRU, 20 August 2026 – The Faculty of Science, Universiti Teknologi Malaysia (UTM), organised Publication Workshop Series 4: Scientific Writing Workshop – Journal Publishing in conjunction with Research Month@FS 2026, with a focus on how researchers can use artificial intelligence responsibly throughout the scientific publication process.
Conducted online via Google Meet, the workshop featured Dr. Ganesan Krishnan from the Department of Physics, Faculty of Science, UTM. His presentation, entitled “Publishing Scientific Papers in the AI Era: What AI Can Do, What It Cannot Do, and How Researchers Can Leverage It Responsibly,” addressed both the opportunities and limitations of integrating AI into contemporary research and manuscript preparation.
A central message of the workshop was that AI is already embedded in many research workflows, including literature exploration, language editing, coding, data analysis, manuscript restructuring and journal-compliance checking. However, the key challenge is not simply whether researchers should use AI, but how to use it without outsourcing scientific responsibility.
Participants were introduced to five recurring principles reflected across major publisher policies: human accountability, prohibition of AI authorship, transparency, verification and confidentiality. Regardless of the AI tool used, authors remain responsible for the accuracy, originality and integrity of their work. Researchers were also reminded to check the latest policies of individual journals, as requirements regarding AI disclosure, image generation, language editing and peer-review confidentiality may differ between publishers.
To help researchers judge appropriate AI use, Dr. Ganesan introduced a practical Green–Amber–Red framework. Green activities include lower-risk applications such as grammar improvement, readability checks, search-term generation and manuscript structure checking. Amber applications, including literature synthesis, suggested references, substantive rewriting, coding, statistics, data visualisation and scientific illustrations, may be useful but require careful verification and disclosure when necessary. Red activities include fabricated data, invented references, AI authorship, alteration of primary evidence and uploading confidential peer-review material to public AI systems.
The workshop then explored the possible role of AI across different sections of a scientific article. Participants were reminded that a scientific manuscript should form one coherent story, moving from why the research was needed, to how it was conducted, what was observed and what the findings mean.
Using an actual final-year project proceedings paper and its related thesis as a case study, Dr. Ganesan demonstrated a controlled approach to AI-assisted writing. The workflow involved examining the original submission, identifying the writing problem, applying AI using clearly constrained prompts, comparing the original and AI-assisted versions, and finally verifying every proposed change before acceptance. This approach emphasised that AI should improve existing work rather than replace the researcher’s scientific judgement.
For the title, AI was demonstrated as a brainstorming partner that could generate alternative titles while preserving key scientific terms and avoiding unsupported novelty claims. For the abstract, AI could function as a completeness checker by identifying whether background, research gap, objectives, methods, results, conclusions and implications were present or missing. Missing information could then be retrieved from the thesis without allowing AI to infer or invent additional details.
A similar approach was applied to the Introduction, where AI served as a structure critic to identify weaknesses in context, established knowledge, research gaps, objectives and claimed contributions. The researcher could then retrieve supporting information from the original thesis and decide how the section should be revised.
For the Methods section, AI was presented as a reproducibility checker. Researchers can use it to identify missing parameters, inconsistent units, insufficient sample-preparation details, absent software information or incomplete statistical procedures. However, AI should never be allowed to guess missing experimental values or create procedural details that were not recorded.
The workshop also highlighted important considerations for figures, graphs and tables. AI can assist in improving presentation quality, generating plotting code, checking labels and units, reformatting tables and developing simple methodology flowcharts. However, researchers were strongly cautioned against using generative AI to alter microscopy images or other primary evidence in ways that add, remove or reinterpret scientific features. The guiding principle was clear: enhance the clarity, not the science.
For the Results and Discussion, AI was demonstrated as a “devil’s advocate” that can challenge overinterpretation, identify unsupported claims, suggest alternative explanations and reveal missing limitations. Nevertheless, decisions about whether an interpretation is scientifically defensible remain the responsibility of the researcher.
AI was also shown to be useful in checking whether a Conclusion adequately answers the stated research objectives using only the reported results. In addition, language editing was presented as one of the safest high-value applications of AI, provided that prompts explicitly preserve scientific terminology, numerical values and causal relationships while prohibiting the addition of new claims or references.
Beyond manuscript writing, the workshop explored the use of AI for reference formatting, journal shortlisting, pre-submission compliance checking, cover-letter preparation and reviewer-response organisation. Researchers were reminded that journal information, indexing, open-access models, fees and current AI policies must still be independently verified before submission.
The session concluded with five practical principles for researchers using AI in scientific publishing: use AI to assist rather than replace judgement; provide verified source material and clear constraints; verify every scientific claim, number, reference and analysis; protect unpublished and confidential material; and check journal policies while disclosing substantive AI use when required.
Through Publication Workshop Series 4, Research Month@FS 2026 provided researchers with practical guidance for navigating scientific publishing in the AI era while preserving research integrity, accountability and scientific quality.
The programme reinforces the Faculty of Science’s commitment to equipping researchers with emerging digital tools while ensuring that the researcher—not the technology—continues to own the science.
#FSResearchMonth2026
#utmfsresearch
#ScientificWriting
#JournalPublishing
#AIinResearch
#ResearchIntegrity
