Using Generative AI in an Undergraduate Thesis | ChatGPT Use, Disclosure, Research Ethics, and Key Precautionsblog

Using Generative AI in an Undergraduate Thesis | ChatGPT Use, Disclosure, Research Ethics, and Key Precautions

The use of generative AI in an undergraduate thesis cannot be judged by a simple "allowed or prohibited" binary. What matters is checking the policies of the university, faculty, seminar, and supervisor and, within the permitted scope, thoroughly observing research ethics, citation rules, personal-information protection, reference verification, and disclosure requirements. This article explains what to check before using generative AI such as ChatGPT in an undergraduate thesis, appropriate use cases, uses to avoid, how to write a disclosure statement, and pre-submission checks.

Using Generative AI in an Undergraduate Thesis | ChatGPT Use, Disclosure, Research Ethics, and Key Precautions

　Generative-AI tools such as ChatGPT, Gemini, Claude, and Copilot can be useful for drafting, summarization, translation, outlining, and organizing search keywords. An undergraduate thesis, however, is an academic work in which the student sets a research topic, reads previous studies, verifies materials or data, and presents the student's own analysis and discussion. AI use therefore requires careful consideration not only of efficiency but also of learning outcomes, research ethics, copyright, information management, and citation rules.

　Japan's Ministry of Education, Culture, Sports, Science and Technology has informed universities and colleges of technology that they should address generative AI in education according to their own educational circumstances and review their guidance as conditions change. Accordingly, whether generative AI may be used in an undergraduate thesis is not governed by one nationwide rule; Give top priority to the instructions of your own university, faculty, seminar, supervisor, and thesis-submission guidelines.the basic principle is:

　Official information from universities such as the University of Tokyo, Kyoto University, Keio University, and Waseda University has also noted that generative AI may support learning while presenting risks such as output errors, bias, leakage of personal or confidential information, academic misconduct, submitting AI output as one's own work, and insufficient disclosure. In an undergraduate thesis, do not accept AI output uncritically; verify sources, exercise your own judgment, and take final responsibility for the work.

What to check before using generative AI in an undergraduate thesis

　Before using generative AI in an undergraduate thesis, first confirm whether your own university permits it. Some universities allow AI use, some require the course instructor's approval, some require the scope of use to be stated, and some prohibit AI in work that must be completed independently. Because an undergraduate thesis has greater implications for research ethics and assessment than an ordinary course report, permission to use AI in a report does not necessarily mean the same use is permitted in a thesis.

　The recommended order of confirmation is: first, the generative-AI policy of the university, faculty, or graduate school; second, the undergraduate-thesis guidelines and submission instructions; third, instructions from the seminar and supervisor; and fourth, requirements relating to research-ethics review and personal-information protection. When handling interviews, questionnaires, experiments, or data from medical, nursing, welfare, or educational settings, entering personal information or unpublished data into an external AI service may be prohibited.

　If generative AI is used, what it was used for, the scope of use, how the output was verified, and who took responsibility for producing the final textit is important to be able to explain the following points. Use that cannot be explained may be judged inappropriate or improper when the undergraduate thesis is submitted.

Key points to understand first

Why university and instructor policies must take priority

　Policies on generative-AI use differ by university, faculty, course, and instructor. MEXT guidance also indicates that universities and colleges of technology should respond independently according to their educational circumstances. It is therefore unsafe to conclude that "AI is convenient, so it must be permitted" or "other students use it, so it must be fine."

　In an undergraduate thesis, topic setting, literature review, research methods, analysis of results, discussion, and conclusion are all part of the assessment. Submitting AI-generated text as your own does not demonstrate your own learning outcomes or research ability and may be treated as academic misconduct. Guidance for students at the University of Tokyo, for example, explicitly states that text or similar output generated by generative-AI tools may not simply be used as one's own writing.

　Keio University's student guidance likewise states that when generative AI is used to prepare reports or similar work, that use must be disclosed, and that AI may not be used for assignments that are required to be completed independently. For an undergraduate thesis as well, it is important to confirm with the instructor how much AI may be used and how the use should be disclosed.

Common risks associated with generative-AI use

　A major problem with generative AI is that it can produce plausible prose quickly even when the content is not correct. Output may contain fabricated references, nonexistent page numbers, incorrect names of laws, invented statistics, misidentified authors, or excessive generalization of research findings. If such material is incorporated into an undergraduate thesis without verification, the reliability of the entire paper is undermined.

　Entering research participants' personal information, interview transcripts, unpublished survey data, or internal materials supplied by companies, hospitals, schools, or local governments into generative AI may also create information-management problems. In studies subject to research-ethics review or confidentiality obligations, careful handling of the information entered into AI systems may be even more important than AI use itself.

Situations in which generative AI can be used more appropriately in an undergraduate thesis

　Generative AI can be useful when treated not as a substitute writer for the undergraduate thesis but as a support tool for organizing research tasks. Examples include organizing perspectives for narrowing a topic, generating candidate keywords for searching previous studies, comparing possible chapter structures, checking readability, explaining technical terminology, producing a preliminary English translation of an abstract, and supporting learning about statistical-analysis procedures.

　AI suggestions should nevertheless be treated only as material for consideration. In an undergraduate thesis, you must decide which literature to adopt, which method to use, which results deserve emphasis, and how the findings should be discussed. Even when you use an AI suggestion, you should be able to explain in your own words why it was accepted or rejected.

Uses that are relatively suitable as assistance

Use to avoid

　One use that should especially be avoided in an undergraduate thesis is having generative AI create the entire body, analysis, discussion, and conclusion and then submitting that output as your own work. This conflicts with the purpose of an undergraduate thesis as evidence of the student's own research ability. Even lightly paraphrasing AI-generated text can still be regarded as ghostwriting or improper use if the student is not the actual intellectual author of the research.

　Research misconduct can also result from listing references whose existence has not been verified, inserting fabricated survey results created by AI, writing as though interviews or questionnaires were conducted when they were not, or presenting unsupported numerical values as facts. Because generative AI can make even false information sound natural, this risk requires particular attention.

Points to note for references and citations

　References are one of the areas requiring the most caution when generative AI is used in an undergraduate thesis. AI can present nonexistent references with convincing-looking bibliographic information. Even when a source is real, the author name, year, paper title, journal name, volume, issue, pages, DOI, or URL may be incorrect.

　Always verify references using CiNii Research, J-STAGE, Google Scholar, university-library OPACs, PubMed, and similar resources. For laws or policy documents, check official pages such as e-Gov Laws and Regulations Search, MEXT, the Ministry of Health, Labour and Welfare, local governments, or universities. Even when AI helps identify information, the final citation should point to a verifiable primary source or academic publication.

How to disclose generative-AI use

　If generative AI is used in an undergraduate thesis, disclosure should follow the requirements of the university, seminar, and supervisor. Even when no format is specified, it is safer to keep a record of the tool name, date of use, purpose, general nature of the input, and how the output was checked and revised so that the process can be explained later.

　Disclosure is not only about whether AI was used. It should clarify what was delegated to AI and what was judged by the student. What is assessed in an undergraduate thesis includes setting the research topic, reading literature, interpreting data or materials, analysis, discussion, and conclusion. Even when AI provides assistance, the final judgment must remain the student's responsibility.

Example wording for an AI-use disclosure

　The following are examples of AI-use disclosure wording. In actual use, always prioritize the required format of the university, seminar, or supervisor. If AI use is prohibited, or the work must be completed independently, these example statements do not make the use permissible.

How generative AI should be positioned in the thesis itself

　Where AI use should be disclosed depends on the submission requirements. If a location is specified, provide the disclosure on the cover page, in a note, methods section, acknowledgments, appendix, AI-use declaration, or other designated place. Even when no location is specified, use in research methods or document organization may be described in the methods section or appendix.

　Generative AI also should not automatically be treated as a conventional "reference." It is not a primary source in the same sense as an academic paper or book, and its output may not be reproducible. Citation and disclosure should therefore follow the university, academic society, or instructor's rules. The key point is that the evidentiary basis of the undergraduate thesis should rest on real literature, materials, data, and research results rather than AI output.

Pre-submission checklist

Support Available from iBooks Academic Support

　iBooks Academic Support assists with undergraduate theses, master's theses, research proposals, university reports, and submitted manuscripts, including organizing concerns about generative-AI use, helping prepare disclosure statements consistent with university policy, verifying the existence of literature, standardizing citation formats, reviewing chapter structure, checking consistency between research purpose and methods, revising prose, and assessing plagiarism and AI-use risks.

　For manuscripts prepared with generative AI in particular, whether the references actually exist, whether the manuscript has drifted into generic AI-generated statements, whether the research purpose and conclusion correspond, whether AI-use disclosure is required, and whether the work conflicts with university or supervisor policyit is important to check these points. We can also assist with pre-submission review of undergraduate theses, responses to supervisor revisions, preparation of AI-use declarations, and renewed verification of references.

Summary | In undergraduate theses, generative-AI use should emphasize assistance and disclosure

　Generative-AI use in an undergraduate thesis cannot be treated as uniformly prohibited or uniformly permitted. The basic approach is to check the policies of the university, faculty, seminar, and supervisor and use AI only as assistance within the permitted scope. Generative AI can help with organizing topics, generating search keywords, comparing chapter structures, revising prose, and supporting translation, but it cannot replace the student's ownership of the research, reading of literature, analysis, discussion, or conclusion.

　The most important point in an undergraduate thesis is to answer your own question by personally verifying the literature and materials and conducting your own analysis. Even when generative AI is used, you must check the accuracy of its output, confirm that references actually exist, protect personal information and unpublished data, and disclose the nature of the use where required.

　If you are unsure whether generative AI may be used in your undergraduate thesis, first check the rules of the submission destination, then consult your supervisor about the permissible scope. The safest approach for protecting the reliability of the thesis is to position AI not as a substitute for research but as a tool for organizing the research process.