Generative AI

Guidance on the use of Generative AI for Graduate Students, their Advisors, and Graduate Programs at the University of Guelph

Institutional background

This resource provides guidance to graduate students, their advisors, and graduate programs on the acceptable and productive use of generative AI in graduate studies. It adds to previously established “Provisional Recommendations for the Use of Generative AI” that focus on how generative AI technologies can and cannot be used in undergraduate and graduate courses.  This resource adds elements of relevance to graduate studies, including the conduct of graduate research, the development of graduate theses, and the completion of Qualifying Examinations.

Generative AI holds opportunities for improving accessibility to research and writing for some students. Departments must support students with accessibility needs; generative AI or tools that incorporate AI might serve as a necessary accommodation. Graduate students, instructors, and advisors with questions or concerns about these accommodations should consult the Graduate Calendar under the Supplementary Academic Accommodation Decision Process and reach out to Student Accessibility Services.

Default Expectation on the Use of Generative AI

Since norms regarding the use of generative AI technologies vary across academic disciplines, graduate students who plan to use generative AI tools in researching or writing, including the completion of the Qualifying Examination, must seek unambiguous approval for planned uses in advance from their advisor(s) and advisory committee.  Approvals must be documented and adhere to departmental and/or field-specific policies where they exist.

Additional Considerations around the Use of Generative AI

In addition to this default expectation, the use of generative AI gives rise to a number of considerations that should be part of any approval offered by an advisory committee.

1. Uses of generative AI should be documented in all outputs that are a part of a submitted thesis.  It is recognized that this documentation may vary from paper to paper within a manuscript-based thesis to conform with journal-specific generative AI policies.
2. Students should be able to interpret and explain all parts of their research that have benefitted from generative AI; no parts of the research can be left "unexplainable".
3. As an author, graduate students are responsible for the accuracy and integrity of their writing, including AI-generated text.  Honest error or lack of intentionality is not a defense in potential research misconduct.
4. Graduate students should understand the consequences of using generative AI tools, and be aware that generative AI:
   1. is likely to generate the most prevalent outcomes - research that is most heavily cited - and alternative theories will be underrepresented;
   2. may include inaccurate or wholly fabricated information, especially in areas of knowledge that are understudied; and
   3. will, through routine engagement, absorb and thereafter make use of graduate students’ outputs without their consent.

For further considerations, please review Artificial Intelligence: Considerations for Graduate Research developed by the Ontario Council on Graduate Studies and published in October 2024.

Going forward

As with our peers, the University of Guelph will continue to think through the implications of Generative AI for our academic and research mission, and adjust expectations as required.  Please check for updates to this resource and look for complementary guidance from the Office of Teaching and Learning and the Office of Research. In addition, academic departments may come to identify norms that fit with their discipline and communicate expectations with graduate students and their advisors.  Ideally, this will manifest in policies and procedures about what is permitted to be produced through generative AI tools for thesis research and outputs.

Acknowledgments

This guidance benefited from leadership from the University of Toronto and the University of Waterloo, coupled with critical initial contributions from a working group consisting of: Karl Cottenie, Associate Dean (Graduate Studies); Katharine Tuerke, CEPS Manager (Graduate Studies); Andrew Bailey, COA Associate Dean (Research and Graduate Studies); and Keriann Mcgoogan, OVC College Research Manager.