Generative Artificial Intelligence

Emily Carr University of Art and Design (ECU) supports the use of Generative Artificial Intelligence to develop a productive learning and working environment in ways that are aligned with our broader commitment to information security, equity, inclusion, accessibility, and sustainability.

Artificial Intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. Generative Artificial Intelligence (GAI) is a subset of AI that focuses on creating new content, such as text, images, and music, by learning from existing data.

In developing our collective GAI literacy, students, staff, and faculty are encouraged to learn how to use GAI tools to enhance their abilities as artists, designers, writers, creative professionals, scholars, and thinkers.

Accordingly, these Guidelines emphasize transparency, fairness, and common-sense limitations for using GAI within our academic community.

Scope

The Guidelines cover all ECU staff, faculty, and students and are meant to address the use of commercially or publicly available GAI tools in relation to their employment duties, research, or coursework.

If you are developing custom GAI tools using ECU devices or resources, you must seek the approval of ECU’s Chief Information Officer prior to commencing any development activities.

Definitions

Context for Guidelines

The Guidelines set out for ECU aim to address the current capacities of GAI, which are evolving, because they are potentially high risk at an individual and institutional level. These Guidelines may be updated regularly to address continued evolution of this technology, alongside a broader effort to establish new policies.

It is also important to acknowledge that there are a range of issues embedded within different GAI platforms including: biases that can cause harm across the many intersectional identities found across our community; use of personal and/or confidential information within these systems that can aid in further training the model and/or used in other generative responses or providing information to other third parties; inaccuracies and inaccurate information (also known as hallucinations); lack of explainability of outputs; and the significant environmental impact of using these systems.

The aim of these Guidelines is to support everyone at ECU in exercising their duty of care and ensuring our community, information, and systems are protected.

Summary of Best Practices

The following list details key best practices for using GAI at ECU:

Information Security and Privacy Considerations

When using a GAI tool, you are prohibited from using any personal information (including sensitive personal information) or confidential information for the purposes of prompting or interacting with GAI tools. Do not share information that you would not want to be made public. This includes personal, sensitive personal, confidential, or proprietary information.

In the case of any potential or actual input of personal information into the GAI tool, the user must contact ECU’s Privacy Officer.

Unless otherwise specified, information you enter into a GAI tool may use your prompts or questions to inform content generated for other users and/or third-parties. It is currently unclear how the material submitted to the third-party is retained, used, or provided to other parties, and how it may inform profit models of the companies creating these tools. Where possible, you should disable the GAI tool’s chat history or model training functions (e.g., OpenAI overview for ChatGPT data controls).

Using GAI in the Classroom

The use of GAI tools is a course-level decision and there is no overall ban on its use in teaching and learning.   

• Instructors should consider course learning outcomes, individual interest, and conventions and expectations of your discipline when determining whether to incorporate GAI tools into course design and activities.
• Instructors should clearly communicate to students if and to what extent generative AI is acceptable in the course outline, verbally in-class and in assessment descriptions.
• Instructors must inform students about these Guidelines for GAI usage when introducing each assignment and advise students that certain uses of GAI technologies may constitute as academic misconduct as per ECU’s Academic Integrity policy.
  
• Instructors must provide clear guidance about attributing GAI tools and consider drafting an example attribution statement as an in-class activity.
• Instructors should consider cost and privacy when designing an assignment where students can use GAI tools. Rely on free versions of generative AI tools for students when possible and provide an alternative for students who do not want to share their information with a third-party tool (e.g., creating a shared account for the whole class).

Sharing students’ work with the GAI tools without their permission can raise a range of privacy and ethical concerns.  A completed assignment is the student’s intellectual property (IP) and should be treated with care.  Students must consent to the sharing of their intellectual work through these GAI tools.

• Instructors should remind students not to input their or anyone else’s creative or intellectual work into GAI tools without understanding how the tool might reuse or store that work.
  

When use of GAI tools is permitted but not required, instructors should ensure fair grading for both those who do and do not use GAI tools by:

• Treating work by students who declare no use of GAI tools as the baseline for grading.
• Using a lower baseline for students who declare use of GAI tools, depending on how extensive the usage, while rewarding creativity, critical nuance, and the correction of inaccuracies or superficial interpretations in response to suggestions made by GAI tools.
• Imposing a penalty for low-energy or unreflective reuse of material generated by GAI tools and assigning zero points for merely reproducing the output from GAI tools.
• Some instructors may prefer stronger restrictions on the use of GAI tools and they are free to impose them so long as care is taken to maintain transparency and fairness in grading.
• When permitted to use GAI tools on assignments, their use should be cited. The American Psychological Association (APA), Modern Language Association (MLA) and the Chicago Manual of Style have all provided recommendations in this area.
• If instructors use GAI in their teaching materials, they should explain in the course outline the extent to which GAI has been, or will be, used. 
• Instructors should also consult any specific TLC resources on GAI when engaging with GAI in the classroom.

Academic Integrity and Use of GAI

A key expectation of academic integrity for students is that they are completing their own work.

• GAI tools should not be used during examinations or assignments, unless explicitly permitted.
• If instructors choose to prohibit GAI tools for an assignment, it is best to suggest that the ‘general use of GAI tools’ is prohibited, as opposed to the use of one particular tool, such as ChatGPT.
• If a student is found to have used a GAI tool for an assignment or exam without explicit permission, the instructor should consider meeting with the student as the first step, in accordance with the University’s Procedures for Cases of Academic Misconduct.

GAI Plagiarism Detecting Software
If an instructor is considering using GAI Plagiarism Detecting software, please note that ECU is still in the process of determining a recommendation for faculty use. In the interim, please consult with the Teaching and Learning Centre before using any software.

Using GAI in Research

Any records classified as research information, including confidential or personal information collected when engaging in research, must not be entered into publicly available GAI tools.

• Confidentiality of research records must be maintained during ECU staff and faculty’s period of employment and in perpetuity, in accordance with the University’s Confidentiality Policy.
• Research projects that rely on GAI as part of the research process must be explicit about their role in the project.

For research projects that require review of a Research Ethics Board application or a Research Data Management Plan, any planned use of GAI tools should be disclosed in the application and project materials, including the informed consent form that details the following:

• The name and details of the GAI tool being used including any publicly available information about the tool that the participant may review;
• Lay person description of the actions performed by the GAI tool and why a GAI tool is needed for the research;
• Overview of the GAI tool’s involvement in the collection, use and disclosure of the participant’s information and which elements of the participant’s personal information will be provided to the GAI tool;
• Whether the GAI tool will use participant information to train the AI model and/or for other business purposes (and whether the information will be de-identified);
• How long the participant information will be retained and how participants can request for their information to be removed and/or destroyed from the GAI tool; and

If any decisions will be made about the participant resulting from outputs of the GAI tool. If so, how the participant can review how the decision was made and refute the decision.

The use of GAI tools throughout the research lifecycle including in the development, editing, or review of grant applications, publications, data analysis, summarizing notes, categorizing results, and creation of novel content must be aligned with all other expectations outlined in these Guidelines, and should be disclosed to the relevant authority in the Research + Industry Office and/or Research Ethics Board.

GAI for Staff Use

Team Leaders should encourage staff to explore how GAI can provide support to their team to complete work more efficiently or effectively.

• Staff who want to use GAI as part of their work must request permission from their manager/team leader and disclose the use of GAI tools to their manager/team leader.
• Team leaders should consult with the CIO on team members’ planned use of GAI tools.
• Staff must review all work and output that is completed or generated through a GAI tool. Submission of unreviewed work should not happen under any circumstances.

All staff must understand how their use of GAI aligns with this document’s Information Security and Privacy Considerations and ECU’s Information Security Standards, Acceptable Use Policy and Confidentiality Policy.

To be clear, this means staff cannot input any personal, confidential, or sensitive personal information related to ECU or its community members into GAI tools.