As Gen AI tools (such as ChatGPT and Dall-E) become increasingly available and widely used, it’s likely that most students will adopt these resources for various academic purposes.  Consequently, it is imperative to help our students develop AI literacy-- including its potential risks-- so that students are equipped with practical AI literacy skills required for future careers. It is equally crucial to emphasize ethical applications, and to remind students that unauthorized use of these tools constitutes academic misconduct, underlining the importance of maintaining integrity when using Gen AI in academic endeavors.

Many possibilities exist regarding the various ways Gen AI can enhance faculty’s courses. Some examples are provided in the 2023 Gen AI Resource Tool. The need to integrate AI for teaching and learning purposes can be both an opportunity and a burden for faculty. To help facilitate this transition, we also recognize the need to specifically include a Gen AI statement in your course syllabi, to clarify when and how use of AI generative tools is permissible, and to what extent. For courses where Gen AI is included in course curriculum, it is expected that both instructors and students understand how to use these resources effectively, appropriately, and ethically, while also remaining mindful of both its usefulness and its limitations. 

The goal of these guidelines is to help faculty foster meaningful learning when Gen AI is included in coursework, while also minimizing opportunities for unethical use. As such, in addition to the university syllabus statement that should be included in all syllabi, we offer the following statement that you might choose to include in your course syllabi. Please note that this is not an official VCU policy; instead it is guidance for a possible statement that you may choose to apply when addressing AI use in your classes.

Suggested Syllabus Statement Regarding Gen AI:
Instructors may choose to provide specific guidelines for use of GEN AI tools for course assignments, exams, or other assessments, or not permit the use of these tools. Instructors must be clear about their expectations including if they are permitting use of AI tools for specific assignment/s for the course, or not at all. If a student believes the appropriateness of use of a particular AI tool is ambiguous or unclear, they must seek clarification from the instructor before using it for an assignment. Instances of using Gen AI-- whether unauthorized, lacking proper documentation, or inadequate citation or attribution-- are considered academic misconduct. This encompasses (but is not limited to) actions such as cheating, deception, inappropriate collaboration, and plagiarism. (See ‘Academic Misconduct’ in VCU’s Honor System and Standards of Academic Conduct for details on how these are defined.) Using Gen AI in a way that violates an instructor’s course policies and/or guidelines violates the VCU Honor System. 

Whether or not faculty plan to incorporate the use of Gen AI into their course, each faculty member should directly address policies about the use of Gen AI in their course syllabus, and provide clear guidance as necessary. For instance, it should be explained if Gen AI is generally permitted within specific guidelines, if it is permitted only under particular circumstances, or if it is prohibited entirely. It is the instructor’s prerogative whether they choose to prohibit its use, allow its use with transparency/acknowledgment by students, or actively include its use within course(s). Additionally, consider that faculty may vary the use of, or deny the use of, Gen AI tools across course assignments/assessments. This should be clearly communicated for each deliverable. 

Remind students that different courses will have different policies, so they should always confirm the specific guidelines for each class with their instructors. We also suggest that faculty review, and refer students to, the Gen AI Guidelines for Students along with the policies/statements contained within their course syllabi at the beginning of each semester.

General Approach to Using Gen AI in Courses 

• AI is not a substitute for teaching. Gen AI should not be used as a substitute for your own teaching. While a useful resource, Gen AI can be used to enhance your teaching, but not replace it.
• Faculty responsibility for teaching materials. As was the case before access to Gen AI, you are 100% responsible for the teaching materials that you use for your course. This continues to be the case if you choose to use Gen AI when developing course content and using it for any particular course materials (e.g., assignments, exams, rubrics, etc.). To establish best practices regarding use of Gen AI, faculty should consider citing/acknowledging their own use of AI when used in their courses.
• Emphasize critical thinking skills. Encourage the use of Gen AI in ways that will inspire critical thinking and creativity of your students, not pose as a substitute for genuine learning. While AI can help support activities that facilitate learning, it is not the process of learning itself as it does not involve active engagement, critical inquiry, and reflection. Gen AI can complement human-centered skills-- such as critical thinking, creativity, and ethical decision-making. Emphasize the importance of these skills in your students’ learning process, for deeper understanding and creative problem solving. 
• Communicate Gen AI policies for your course(s). Make clear what your policies are regarding Gen AI use in your course. Expectations for ethical and responsible use should always be emphasized, as well as modeled by the instructor. Lead by example, by demonstrating best practices when developing your course. If you have used Gen AI when creating certain course materials, acknowledge its use.
• Gauge quality against efficiency.  Gen AI can help increase your efficiency, such as with creating and grading assignments. However, faculty should critically assess the overall quality of what has been created with the assistance of Gen AI, to determine if its output is suitably aligned with your curriculum.
• Always consider data privacy. Remind students that, when using publicly available AI resources, entering personal data or other confidential information is a security risk. AI systems cannot guarantee data privacy, and currently there are no standardized security protocols for these systems. (See additional information from the VCU Office of the CIO, regarding protection of confidential data.)
• Protecting student data. In addition to student information (such as  email addresses) all work submitted by students is protected under FERPA, and therefore cannot be uploaded to a Gen AI tool without their written consent. (This includes Gen AI detection tools, unless VCU has a contract with the provider of that particular detection tool.) Students, however, can upload their own graded materials into such tools. This, however, is not recommended unless VCU has a contract with a Gen AI tool’s provider, specifying that the submitted data will be protected.
• Support course objectives. Always ensure that your Gen AI resources/materials support your course objectives/outcomes. Regularly assess the impact of using AI in your course and student learning outcomes. Solicit feedback on course activities that use AI, and adjust when needed to optimize learning and skill development.
• Promote bias-free resources. Confirm that Gen AI resources that you suggest or require for your courses are not biased, prejudiced, or inaccurate in the information they produce.
• Confirm accessibility for students. Verify if the Gen AI resources are accessible and freely available for all students, to ensure equitable access.  Ensure that AI usage does not disadvantage any students, and promotes inclusivity for diverse learning preferences and needs. (VCU IT Governance provides additional information, here.)
• Remain up-to-date. Keep yourself informed regarding development of new AI resources, ethical considerations, and ways to potentially integrate Gen AI into your curriculum.

Generative artificial intelligence tools are rapidly transforming our teaching and learning practices at VCU. These tools will continue to shift our approach to both regular tasks and how we address new challenges. The CTLE advises faculty to reflect on the ethical considerations of using Gen AI both in and outside of the classroom. 

• Emphasize ethical use of Gen AI. You should confirm that Gen AI resources you use are free from bias, prejudice, and inaccuracies in the information they produce, while also adhering to ethical standards. This includes respecting privacy, confidentiality, and intellectual property rights, and using data in a responsible manner that complies with relevant regulations.
• Intellectual property concerns. Using Gen AI raises many questions about intellectual property. For instance, responses or content generated by AI may lead to concerns about citing, copyright, or ownership. Currently, ownership of content generated by Gen AI remains unresolved at the federal level, and there is no clear standard regarding its ownership.
• Student data is protected. All graded work done by students is protected under FERPA, therefore it should not be uploaded to a Gen AI tool without students’ written permission.
• Confirm accessibility of Gen AI resources for students. Gen AI resources used in courses should be accessible and freely available for all students, to ensure equitable access. Before allowing/assigning use of Gen AI resources, always confirm their accessibility for students.
• Performance limitations. Among various important limitations, as noted in a report by Open AI (2023), Gen AI does not perform well with multiple elements such as providing (accurate) citations, knowing current events, multi-step math problems, along with being, “confidently wrong in its predictions.” In fact, when ChatGPT itself was prompted about its limitations at the time of this write-up, it listed current events, contextual understanding, critical thinking, emotional understanding, and factual accuracy as potentially problematic.  
• Process of Generating Answers is Unclear. While Gen AI provides output to queries, the internal processes it uses to generate content are still unclear.
• Bias and quality of data. Gen AI output is based on existing data, including data taken from online sources, as well as data input when training. This may result in output that reflects biases, inaccuracies, or complete fabrication. ChatGPT, itself, provides various disclaimers such as:
  “While we have safeguards in place, the system may occasionally generate incorrect or misleading information and produce offensive or biased content. It is not intended to give advice.”
• Intellectual property concerns. Using Gen AI raises many questions about intellectual property. For instance, responses or content generated by AI may lead to concerns about citations/attribution, copyright, or ownership. Currently, ownership of content generated by Gen AI remains unresolved at the federal level, and there is no clear standard regarding its ownership.
• Transparency regarding responses. While Gen AI provides output to queries, the internal processes it uses to generate content are still unclear.
• Concerns with plagiarism detection tools. Currently, the accuracy of tools that purportedly detect GenAI use is not guaranteed. There are instances where these tools fail to detect GenAI use, may not accurately distinguish between content that is generated by humans versus generated by AI, and may generate false positive or negative results. These tools may also exhibit inherent biases that can negatively impact certain groups of students (e.g., non-native English writers, etc.), creating issues of fairness. (See, for example, Liang et al. 2023.)

Initial drafts and extensive edits of the guidelines were created by Dr. Susan Coombes, Dr. Oscar Keyes, and Dr. Mangala Subramaniam. Additional edits and suggestions contributing to the current version of the guidelines were also provided by Jake Belue, Dr. Andrew Arroyo, Dr. Jonathan Becker, Professor TyRuben Ellingson, Dr. Karen Belanger, Kristin Calleja, Sarah Johns, Dr. Rochelle Jordan, Dr. Elaine Reeder, Dr. Jennifer Rhee, Dr. Kate Sicchio. The input and guidance of all contributors is greatly appreciated.