How do you use AI ethically in your literature review?

Using AI ethically in your literature review means using it as a search and screening assistant, never as a source of citations you do not verify.

Using AI ethically in your literature review means using it as a search and screening assistant, never as a source of citations you do not verify. For Vietnamese researchers under pressure to publish in Scopus Q1 and Q2 journals, the line that protects your integrity is simple: AI can help you find and organise the literature, but you remain accountable for every claim, every reference, and every word that reaches an editor.

This guide answers the seven questions Vietnamese researchers ask MAAS publishing mentors most often about using AI in the literature review stage without crossing journal or university rules.

Author: MAAS AI & Data Science Publishing Desk · Reviewed by a Principal Publishing Advisor (PhD, Scopus Q1 author and reviewer)
Last updated: 2026-06-05
Category: research-methods

What does it mean to use AI ethically in a literature review?

Direct answer: Using AI ethically means letting it accelerate tasks you can still verify and take responsibility for — searching, deduplicating, screening, and summarising for your own understanding — while never delegating judgement, citation, or authorship to the tool. The principle that ties every rule together is human oversight: you justify the use, you check the output, and you own the result.

Evidence: A 2025 joint position statement from Cochrane, the Campbell Collaboration, JBI, and the Collaboration for Environmental Evidence concluded that current evidence does not support generative AI use in evidence synthesis without human involvement, and that any AI use must be disclosed, kept under human oversight, and justified so it does not compromise methodological rigour (Cochrane Library, 2025).

Example: A Vietnamese doctoral candidate MAAS coached used an AI tool to cluster 600 search results by theme. The mentor's rule was that the AI could group and rank papers, but the candidate had to read and confirm every included study herself — so the screening was faster, and the accountability stayed human.

Can AI tools be listed as authors or sources in your paper?

Direct answer: No. AI tools cannot be authors, because they cannot take responsibility for the work, declare conflicts of interest, or approve a final version. They also are not citable sources for factual claims. Treat any AI output as a draft thought to verify against real, peer-reviewed literature, not as evidence in itself.

Evidence: The COPE position statement (2023) states that AI tools cannot be listed as authors under any circumstances because they cannot meet authorship criteria or be accountable. The ICMJE recommendations likewise hold that chatbots cannot be authors and that humans remain responsible for all material produced with AI assistance (ICMJE, updated 2026).

Example: A MAAS-coached researcher had pasted a ChatGPT-generated paragraph, complete with citations, into her introduction. Her mentor traced the references and found two did not exist. They removed the passage, rebuilt the argument from sources she had actually read, and the introduction became both defensible and original.

Why shouldn't you trust AI for citations and summaries?

Direct answer: Because general-purpose chatbots fabricate references that look real — plausible authors, journals, and even correctly formatted DOIs — for studies that were never written. Summaries can also misstate a paper's findings. Every citation an AI suggests must be located and read in the original before it enters your review.

Evidence: A study in Nature Scientific Reports (2023) found that around 55% of bibliographic citations generated by GPT-3.5 were fabricated and many real ones contained errors; newer models reduced but did not eliminate the problem, with later analyses still finding roughly one in five AI-generated references entirely fake (2024–2025). The fabricated citations were deliberately hard to spot because they mimicked legitimate formatting.

Example: A Vietnamese master's student asked an AI tool to "find five papers on her topic." All five were convincing and all five were invented. Her MAAS mentor showed her how to verify each reference in Scopus and Google Scholar before use — a five-minute habit that prevents a fatal credibility error at review.

Which AI tools are appropriate for the literature review stage?

Direct answer: Favour tools built for academic search and screening, where every result links back to a real, locatable paper, over open-ended chatbots that generate prose. Discovery and screening tools surface genuine literature and save time; generative chatbots are best limited to language help on text you wrote, with disclosure.

Evidence: Elicit links claims to sentence-level citations in real papers; Semantic Scholar, run by the non-profit Allen Institute for AI, indexes over 200 million papers with semantic search. For screening, Rayyan uses active learning and survey respondents reported around 40% time savings, while the open-source ASReview offers transparent, reproducible prioritisation of records (Springer Systematic Reviews, 2016; PMC, 2023–2025).

AI task in a literature review	Appropriate use	Tool type / example
Finding real papers	Yes — verify each result	Discovery search (Semantic Scholar, Elicit)
Screening titles/abstracts	Yes — human confirms every decision	Active-learning screeners (Rayyan, ASReview)
Summarising a paper you have	Yes — for your own understanding, then read it	Source-grounded summariser
Generating citations or "facts"	No — high fabrication risk	General chatbot (avoid for this)
Light language polishing of your text	With disclosure and oversight	General writing assistant

Example: A MAAS-coached health-sciences author replaced "ask ChatGPT for sources" with a workflow of Semantic Scholar for discovery and Rayyan for screening. Her search became reproducible — she could show exactly how papers were found and selected — which is itself a quality signal Q1 reviewers value.

Do you have to disclose AI use, and where?

Direct answer: Yes, whenever AI materially assisted your work. Disclose at submission, name the tool and how you used it, and place the statement where the journal directs — usually the acknowledgements for writing help and the methods section for screening, data, or analysis. Routine grammar checking is often exempt, but disclosing is the safe default.

Evidence: The ICMJE recommends that authors disclose at submission whether AI-assisted technologies were used, describing the use in the cover letter and in the appropriate section — acknowledgements for writing assistance, methods for data work (ICMJE, updated 2026). The RAISE collection from Cochrane (2025) similarly requires that AI and automation in evidence synthesis be disclosed and justified.

Example: A Vietnamese researcher used ASReview to prioritise screening for a scoping review. Her MAAS mentor helped her write a two-sentence methods disclosure stating the tool, version, and how human reviewers checked its rankings — turning a compliance requirement into evidence of a rigorous, transparent process.

Is it safe to upload papers or your manuscript to an AI tool?

Direct answer: Be very careful. Uploading an unpublished manuscript, a confidential dataset, or a paper you are peer-reviewing to a public AI tool can breach confidentiality, intellectual-property, and data-privacy rules — even when a tool claims to be "private." For sensitive material, assume anything you upload may be stored or reused.

Evidence: Elsevier's policy states that submitted manuscripts are confidential documents, and uploading a manuscript or any part of it to an AI tool may infringe authors' confidentiality and IP rights, because many tools retain or reuse uploaded content (Elsevier, 2024–2025). The same logic applies to manuscripts you are reviewing for a journal.

Example: A MAAS-coached author wanted to "let AI summarise" a confidential collaborator draft. Her mentor flagged the IP risk, and instead they worked from her own notes and published sources. The review stayed strong and no unpublished work left her control.

How can Vietnamese and ESL researchers use AI without crossing the line?

Direct answer: Use AI to reduce the mechanical load of searching, organising, and language polishing, while keeping the intellectual work — reading, judging, citing, and arguing — firmly yours. For ESL researchers, AI language help is legitimate when it edits text you authored and you disclose it; it becomes a problem when it generates claims or sources you did not verify.

Evidence: Vietnam's national science strategy targets a 15–20% annual rise in WoS/Scopus/Q1 output and ties doctoral progression to international publication (Vietnam national science program, 2024–2025), raising the stakes for getting AI use right. Reporting guidelines and publisher policies converge on the same rule: transparency plus human accountability keeps AI use ethical (COPE 2023; ICMJE 2026; Cochrane 2025).

Example: A MAAS mentor coached a Vietnamese researcher through the Outline → Draft → Final model with AI kept in its lane: AI-assisted discovery and screening in the outline stage, the candidate's own analysis in the draft, and disclosed language polishing at the final stage. She stayed the author throughout, with the mentor advising — never writing — at each step.

Frequently asked questions

Is it cheating to use AI in a literature review?
Not by itself. Using AI to find, screen, or organise real literature, with disclosure and verification, is legitimate at most universities and journals. It becomes misconduct when you present AI-generated text or unverified citations as your own verified work, or hide the use when asked to declare it.

Can Turnitin or journals detect AI-generated text?
Detection tools exist but are imperfect and produce false positives, so do not rely on "passing" them. The real safeguard is doing the work properly: write your own analysis, verify every source, and disclose AI assistance. Integrity, not evasion, is the standard editors apply.

Which AI tool is best for finding sources?
Prefer source-grounded academic tools such as Semantic Scholar or Elicit, where every result links to a real, locatable paper, over open chatbots that can invent references. Whatever the tool, confirm each paper in a database like Scopus before citing it.

Do I need to disclose AI use if I only used it for grammar?
Routine grammar and spell-checking is often exempt, but policies vary by journal. When in doubt, add a short disclosure — it costs nothing and protects you. Substantive language rewriting or content generation should always be disclosed.

Can AI write my literature review for me?
No — and you should not want it to. A literature review is your synthesis and argument, which reviewers expect to come from you. AI can support search, screening, and language, but the critical reading and reasoning must be yours to withstand peer review and to be ethical.

Can MAAS help me use AI responsibly in my research?
Yes. MAAS Publishing Advisory coaches Vietnamese researchers on integrating AI search and screening tools, disclosure statements, and methodology that meets journal and integrity standards, using the Outline → Draft → Final model. Book a consultation through our contact page.

Ready to build an AI-assisted review that survives peer review?

Used well, AI makes your literature review faster and more reproducible; used carelessly, it introduces fabricated citations and integrity risks that can sink a submission. MAAS Publishing Advisory pairs you with a PhD-level mentor — 23% of our experts hold doctorates — for a free 20-minute consultation, matches you to the right advisor within 48 hours, and backs every engagement with our three-tier Pass / Merit / Distinction guarantee and a 90-day post-submission warranty. We coach; you stay the author, every step.

Book a Publishing Advisory consultation with MAAS Academic Mentoring →

How do you design a systematic review in the health sciences? — where AI screening tools fit a rigorous review protocol
How do you publish a medical imaging AI study in a Q1 journal? — reporting and transparency standards for AI-in-health work
How do you avoid predatory journals when publishing your research? — verifying sources and venues before you commit
How do you choose the right Scopus journal for your paper? — matching your verified review to the right target journal
Scopus publishing resources — the full MAAS hub for Q1/Q2 publishing support
Publishing Advisory service — mentor-led support across feasibility, methodology, and submission
Meet the MAAS experts — the PhD-level mentors behind our publishing advisory

References

This article is part of the MAAS Journal series for Vietnamese international postgraduate students and researchers. MAAS Publishing Advisory is an advisory partner — we coach authors through the Outline → Draft → Final delivery model with developmental feedback from PhD-level, Scopus-published mentors. We do not write, submit, or guarantee acceptance of work on an author's behalf.