Provisional Recommendations for the Use of Generative AI
Provisional Recommendations for the Use of Generative Artificial Intelligence (AI) in Teaching and Learning at the University of Guelph
Provisional Recommendations for the Use of Generative Artificial Intelligence (AI) in Teaching and Learning at the University of Guelph – July, 2023 by the University of Guelph is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
The Office of Teaching and Learning (OTL) at the University of Guelph (UofG) has been engaged in cross-institutional conversation in an initiative led by the MacPherson Institute at McMaster University, to develop guidelines for the use of generative artificial intelligence (AI) in teaching and learning for instructors.
As a starting point for UofG instructors to consider the potential uses of generative AI in teaching and learning, we have adapted the following recommendations from provisional guidelines developed by McMaster’s Task Force on Generative AI in Teaching and Learning through a Creative Commons Attribution-NonCommercial 4.0 International License.
OTL intends to continue dialogues across units, departments, Colleges, Guelph-Humber and Ridgetown campuses, as well as between post-secondary institutions, to refine and expand upon these recommendations as new information and technology emerges and evolves, and as we continue to learn to work with these technologies. These recommendations are based on the current understanding and institutional policies that exist related to generative AI technologies, as well as the technology itself. We recognize that these can rapidly change, and the recommendations will be updated accordingly. Institutional policies take precedence over these recommendations.
The recommendations below recognize that:
- the use of generative AI will differ based on disciplinary cultures
- individuals will hold different expectations and reactions to the technology
- importantly, such technologies are accompanied by significant ethical debates, inequitable access, and may reproduce harmful biases (see for example issues raised in Rebecca Sweetman and Yasmine Djerbal, “ChatGPT? We need to talk about LLMs,” University Affairs, 25 May 2023)
- and there are also optimistic outlooks of generative AI’s potential in higher education (see for example Bill Gates, “The Age of AI has begun,” GatesNotes: The Blog of Bill Gates, 21 March 2023, especially the section on Education; and Ethan Mollick, “Democratizing the future of education,” One Useful Thing, 21 April 2023).
In creating these provisional recommendations, we gratefully acknowledge the feedback provided by instructors and staff from various units and Colleges across the UofG. OTL welcomes additional feedback and requests for consultation from instructors regarding these recommendations. Please e-mail us at email@example.com for support.
Check back on our Teaching in the Context of AI page for other supports that OTL will offer, including upcoming sessions.
Within the next six months, OTL will review and update these provisional recommendations according to new information and feedback. This current version was last updated on 29 August 2023.
Recommendations for Instructors to consider when incorporating generative AI for their own use in course planning and design
Instructors should review the University of Guelph-issued statement on artificial intelligence systems, ChatGPT, and academic integrity from March 2023.
It is up to each individual instructor to decide if they want to incorporate generative AI into their courses’ design, activities, and assessments in creating course content or for student use. If an instructor chooses to incorporate generative AI, it should be based on course learning outcomes, instructors’ individual interests, and conventions and expectations of the discipline. Instructors are not required to use generative AI tools for teaching and learning.
As with any pedagogical tool or approach, instructors should weigh the benefits of incorporating generative AI into their courses against any risks inherent to the tool or approach. Instructors should also take into account the rapidly evolving nature of generative AI technology and reassess the benefits and risks of any tool or approach on a regular basis.
Instructors with courses that incorporate generative AI for student use should:
a) Build AI literacy for their own benefit, and open conversations with students, by learning more about artificial intelligence in higher education. Consider using our resources to build AI literacy and to facilitate discussions of generative AI for student use.
b) Consider the course learning outcomes and ensure the incorporation of generative AI for student use will support core learning outcomes; and ensure incorporation offers meaningful learning, or experimentation. Its use should be intentionally incorporated into the course design, rather than inclusion for the sake of novelty.
c) Describe or discuss with students the strengths, limitations, and ethical considerations of the technology, including factual inaccuracies or ‘hallucinations’, societal biases including racism and sexism, present in the training data and the rationale for using generative AI in assignments.
Until standards and/or policies have been established for instructors on their use of AI for instructional materials, instructors should be transparent and explain in the course outline the extent to which generative AI has been, will be, or might be used. Instructors should model responsible use of generative AI as per the recommendations in this document.
Any materials produced by generative AI for instructional use should be fact-checked by instructors, unless the outcome of its use is to demonstrate the limitations of generative AI produced content.
Recommendations for Instructors to consider when incorporating generative AI into courses for student use
Individual instructors should clearly communicate to students if and to what extent generative AI use by students is acceptable in their course. This should be communicated in the course outline, in the learning management system (e.g., CourseLink), verbally in-class, and in assessment descriptions. A helpful starting point is OTL and McLaughlin Library’s Tool for Determining Allowable Uses of AI with Writing Assignments, a template that can be adapted for specific courses. See Appendix B below for further examples of how this could be included in course outlines.
Instructors who include assessments that incorporate generative AI for student use should:
a) Seek to foster an environment of academic integrity when creating assessments, including with generative AI. Click here for a PDF on overall best practices on fostering an environment of academic integrity in assessment.
b) Consider assessment types that may be less susceptible to the use of generative AI such as authentic assessments, oral exams, presentations followed by a Q and A, invigilated/in-class assessments, practical tests, assessments that incorporate class discussion/activities or critical reflection, and process-based work, such as annotating source materials to show depth of understanding.
c) Consider including critically reflective components that invite students to comment on the use of/experience with generative AI in the assessment.
d) Explicitly review criteria and/or rubrics in ways that demonstrate how the use of generative AI is being assessed. See Appendix C below for example.
e) Explain to students in writing and verbally in-class how generative AI material should be acknowledged or cited. If students use generative AI, they should consult with the relevant McLaughlin Library’s citation guides for MLA, APA, and Chicago on how to cite ChatGPT. See Appendix A below for examples.
f) Be aware of the privacy policies and user agreements of instructor-selected AI tool and alert students to these policies in the course outline, or in the description of assignments or activities involving generative AI. Please feel free to consult with OTL if you seek guidance on creating a course outline statement to this effect.
g) Keep in mind privacy policies, noting that where possible, courses that incorporate generative AI should rely on free versions of generative AI tools (e.g., Microsoft Bing, ChatGPT 3.5) for student use. Alternatives assessments should be provided for Generative AI tools that are restricted to users 18+ (e.g., ChatGPT), and in cases where good faith concerns relative to privacy or other terms and conditions of use are raised.
h) Before submitting student work to generative AI tools, consider the risks involved, particularly when using tools not approved by the UofG. Submission of student work to generative AI tools may constitute an infringement upon students’ privacy or intellectual property rights. Given that generative AI is a rapidly evolving area, you are invited to converse with OTL about the risks that need to be considered. OTL is actively collaborating with university units such as the University Secretariat and CARE-AI to better understand the effects of generative AI on teaching, learning, and privacy. Such collaborations are intended to guide individual instructors, departments, and Colleges, to benefit our teaching and learning community in adapting to generative AI. Also see recommendation 8 below for specific considerations on submitting student work to AI plagiarism detection software.
Generative AI plagiarism detection software is not currently approved for use at UofG. This software will continue to be reviewed and may be available in the future.
a) Many of these detectors will produce false positives or negatives, and their use may constitute an infringement upon the students’ privacy or intellectual property rights. In many cases, it is unclear how the material submitted to third-party detectors is retained or used.
b) Until more is understood about generative AI detection tools, instructors should not submit student work to generative AI detection tools. (Please note that this is different from submitting student work to generative AI tools – for example, to get AI-generated feedback – as noted in recommendation 7(h) above.)
c) UofG has an institutional subscription to Turnitin, a plagiarism detection software. Turnitin announced an update aimed at detecting writing produced by generative AI, which has brought about a variety of concerns. UofG, like many other institutions, has not yet turned on this feature as there is a need to understand the functionality of the tool, assess the security and privacy considerations for student work, and determine whether it aligns with existing policies.
d) If you do suspect student work may have violated the academic integrity policy, please refer to the relevant academic misconduct sections in the undergraduate, graduate, and associate diploma calendars.
If the style guide you use in your discipline has yet to release guidelines, instructors and students should develop a citation option that clearly cites use of generative AI, such as:
- “[Generative AI tool]. (YYYY/MM/DD of prompt). “Text of prompt”. Generated using [Name of Tool.] Website of tool”
- e.g., “ChatGPT4. (2023/05/31). “Suggest a cookie recipe that combines oatmeal, chocolates chips, eggs and sugar.” Generated using OpenAI’s ChatGPT. https://chat.opeani.com”
As per recommendations 7(b) and 7(c), instructors may also consider requiring students to include a reflective summary at the end of each assessment that documents what generative AI tools were used, what prompts were used – including a complete chat log – and how generated content was evaluated and incorporated. In addition to citation, instructors may also consider having students verify the information generated by AI.
First, refer to Recommendation 6 above. Consider using OTL and McLaughlin Library’s Tool for Determining Allowable Uses of AI with Writing Assignments, a template that can be adapted for specific courses. OTL welcomes instructors to submit and share other assessment- and discipline-specific tools to guide instructors, including adapting this template.
Following are examples of UofG Course Outline Statements and/or Assignment Instructions that instructors can use to prohibit, restrict, or allow students to use generative AI in their courses:
Students are not permitted to use generative AI in this course. Please refer to the University of Guelph-issued statement on artificial intelligence systems, ChatGPT, and academic integrity from March 2023 and regulations and procedures around academic misconduct in the undergraduate and graduate calendars.
Some Use Permitted
Students may use generative AI in this course in accordance with the guidelines outlined for each assessment, and so long as the use of generative AI is referenced and cited following citation instructions given in the course outline and/or assignment instructions. Use of generative AI outside assessment guidelines or without citation will constitute academic misconduct. It is the student’s responsibility to be clear on the limitations for use for each assessment and to be clear on the expectations for citation and reference and to do so appropriately.
Students may use generative AI for [composing, editing, translating, outlining, brainstorming, revising, etc.] their work throughout the course so long as the use of generative AI is referenced and cited following citation instructions given in the course outline and/or assignment instructions. Use of generative AI outside the stated use of [composing, editing, translating, outlining, brainstorming, revising, etc.] without citation will constitute academic misconduct. It is the student’s responsibility to be clear on the limitations for use and to be clear on the expectations for citation and reference and to do so appropriately.
Students may freely use generative AI in this course so long as the use of generative AI is referenced and cited following citation instructions given in the course outline and/or assignment instructions. Use of generative AI outside assessment guidelines or without citation will constitute academic misconduct. It is the student’s responsibility to be clear on the expectations for citation and reference and to do so appropriately.
Students may use generative AI throughout this course in whatever way enhances their learning; no special documentation or citation is required.
The following are sample rubrics generated by McMaster University. Note recommendation 3(a) on the importance of building AI literacy prior to creating assessments using generative AI. OTL welcomes feedback and input from UofG instructors to refine these and offer further examples:
McMaster University acknowledges the assistance of ChatGPT 4.0 to create these sample analytic and holistic rubrics. The prompts included “Imagine you are a rubric generating robot who creates reliable and valid rubrics to assess university-level critical thinking skills. You have been tasked with generating a rubric that evaluates students critical thinking skills and incorporates their use of generative AI. Create two holistic rubrics and two analytic rubrics to assess these skills.” With human editing, the output from these prompts was to provide examples of the kind of rubrics that could be used to assess the integration of generative AI in course assignments.
Analytic Rubric 1: Assessing Reflection on Generative AI Use
|The argument is clearly articulated and logically structured.
|The argument is generally clear and logical, with minor inconsistencies.
|The argument is somewhat unclear or inconsistently structured.
|The argument lacks clarity and logical structure.
|Evidence is thorough, relevant, and convincingly supports the argument.
|Evidence is generally strong and relevant, with minor lapses.
|Evidence is somewhat sparse, irrelevant, or does not fully support the argument.
|Evidence is lacking or largely irrelevant.
|Use of Generative AI
|AI is used effectively to support arguments, demonstrating a high understanding of its capabilities and limitations.
|AI is used effectively, but understanding or integration could be improved.
|AI is used, but not effectively integrated or misunderstood.
|AI is not used or its use does not contribute to the argument.
|Reflection on AI
|The student clearly articulates how AI contributed to their critical thinking process and considers its limitations.
|The student generally explains how AI contributed to their thinking, with minor lapses in considering its limitations.
|The student’s explanation of how AI contributed to their thinking is unclear or superficial.
|The student does not explain how AI contributed to their thinking.
Analytic Rubric 2: Assessing Generative AI Use and Integration
|Understanding of AI
|The student demonstrates a deep understanding of the capabilities and limitations of the AI.
|The student demonstrates a good understanding of the AI, with minor misconceptions.
|The student demonstrates a good understanding of the AI, with minor misconceptions.
|The student shows little to no understanding of the AI.
|Integration of AI
|AI is seamlessly integrated into the work, effectively augmenting the student’s critical thinking.
|AI is generally well integrated, though at times it may seem somewhat forced or awkward.
|AI integration is inconsistent or superficial, not effectively augmenting the critical thinking process.
|AI is not effectively integrated into the work.
|Reflection on AI
|The student clearly reflects on the role of AI in their work, considering both its contributions and its limitations.
|The student generally reflects well on the AI’s role, though considerations of its limitations may be superficial.
|The student’s reflection on the AI’s role is minimal or lacks depth.
|The student does not reflect on the AI’s role in their work.
|Innovation with AI
|The student uses AI in novel or innovative ways to enhance their argument.
|The student uses AI effectively, though it may lack innovation.
|The student uses AI in a straightforward or predictable way, not enhancing the argument.
|The student does not use AI in an innovative or meaningful way.