GW Guidance and Expectations: Use of GAI in Teaching and Learning

This document reviews guidance related to the institutional policies affecting faculty use of Generative Artificial Intelligence (GAI) in teaching, provided by the GW Provost and applicable GW offices. Faculty use of GAI should also comply with federal and local laws governing higher education. Key policy and guidance are summarized below, followed by key considerations based on that guidance.

GW Provost’s Office Guidance

In April 2023, the Provost disseminated guidelines with an update in August 2023 that included the following key requirements and suggestions applicable to faculty, excerpted below.

“The Office of the Provost encourages the entire University community to embrace these technologies through creative uses and applications. Faculty are invited to make thoughtful use of GAI tools in their teaching and research… Even as we are learning ourselves, we must teach our students to use GAI tools effectively and responsibly.”

“The Office of the Provost encourages instructors to state explicitly and affirmatively their expectations regarding student use of GAI tools. Instructors should specify in writing the permitted and prohibited uses of GAI tools in their courses. Instructors might 1) generally permit the use of GAI tools; 2) generally forbid their use; or 3) permit their use for certain purposes on certain assignments, but not others. If an instructor wishes to permit certain uses of GAI tools, such uses must be set forth explicitly in the course syllabus and/or assignment instructions.”

“In the absence of explicit directions to the contrary from instructors, the following default rules apply at the University:

Work submitted for evaluation is represented as the student’s own intellectual product. Students may not submit content (e.g., ideas, text, code, images) for evaluation that was generated, in whole or in part, by Generative Artificial Intelligence tools (such as ChatGPT and other large language models). Doing so without the instructor’s explicit permission constitutes cheating under the Code of Academic Integrity¹” and is therefore prohibited” [and can be reported using the same process as any other violation of academic integrity].
Students are permitted to use GAI tools to generate content that is not submitted to an instructor for evaluation. For example, using GAI tools to study for examinations, tests, and quizzes is permitted. Likewise, on assignments where the use of the Internet is not otherwise prohibited by the instructor, GAI tools may be used for learning, studying, and brainstorming.
Unless the instructor explicitly states otherwise in advance and in writing, the use of GAI tools during any assessment (e.g., examination, test, quiz), whether taken in the classroom or elsewhere, constitutes cheating under the Code of Academic Integrity and is therefore prohibited. This prohibition includes assessments for which the use of the Internet is otherwise permitted.”

At this time, there is no university policy on whether students can be required to use GAI. In the meantime, it is recommended that faculty create classroom environments that allow students to engage in learning regardless of the student’s position on GAI.

¹Academic Integrity Policy (applies to all but Law, Medicine, and Pre-College students), Section 1(a): Definition of Academic Integrity Violations – Academic integrity violations are cheating of any kind, including misrepresenting one's own work, taking credit for the work of others without crediting them and without appropriate authorization, and the fabrication the of information.

GW’s Privacy Office Guidance

The GW’s Privacy Office has issued guidance for the use of AI. Although primarily related to administrative uses of AI, it also states the following relevant to teaching:

“Staff and faculty must be vigilant about the data they enter into any AI tools, and must be certain that the use of AI tools will not, in any way, violate GW’s privacy, data protection and security policies and requirements.
Before using an AI tool that accesses personal information of individuals, users should carefully consider whether it is necessary to include this information and whether the desired outcome can be achieved without personal details.
Maximum privacy and security controls must always be employed to protect personally identifiable information (PII). Accordingly, when PII is entered into an AI tool, the AI tool should be set with privacy-enhancing settings to ensure that the PII is not retained by the AI Model and also, when possible, the PII should be anonymized.
AI tools used at GW must not use personal information of our community members for training their models.”

Relevant Federal and Local Laws

By shifting how we work, teach, and learn, AI also creates new ways to violate longstanding student rights. For this reason, although faculty may use AI in a range of ways to support learning, it should not be used at the expense of student privacy, access, equity, dignity, or human judgment. The major relevant federal and local laws are discussed below:

FERPA (Family Educational Rights and Privacy Act)

FERPA (Family Educational Rights and Privacy Act) is a federal privacy law that protects the confidentiality of student education records and limits how student-identifiable information may be shared or disclosed. At GW, this law is overseen by the GW Privacy Office. Applying FERPA to GAI, faculty should avoid providing student-identifiable information (grades, records, identifiable coursework) to any GAI system unless the university has formally approved the tool for FERPA compliance.

Section 504 + ADA (Americans with Disabilities Act)

Section 504 + ADA (Americans with Disabilities Act) are disability laws that require institutions to provide equal access to education and prohibit discrimination against students with disabilities, including through the denial of reasonable accommodations and assistive technologies. At GW, these laws are overseen by the Office of Disability Support Services (DSS). Approved disability accommodations apply to all classwork course components. Faculty may not ban approved assistive technologies that function as accommodations, although if a faculty member is concerned that an accommodation fundamentally alters an essential element of the course/assignment, they should request a review by DSS. Further, no student may be required to disclose disability status.

Title VI + DC Human Rights Act

Title VI + DC Human Rights Act: Title VI is a federal civil-rights law that prohibits discrimination based on race, color, or national origin, while the D.C. Human Rights Act is a local civil-rights law that provides broader protections for additional protected characteristics. Implementation of these laws at GW is overseen by the Office of Access and Opportunity. When using GAI in the classroom, faculty should do so in a way that does not create differential educational access. GAI use must be optional or supported by equivalent free alternatives. AI may not be used for surveillance or automated decision-making that affects students’ standing or opportunity.

Title IX

Title IX is a federal law that prohibits sex-based discrimination and harassment in educational programs and requires institutions to respond promptly and effectively to related misconduct. This law is overseen by GW’s Title IX Office. AI systems may not be used to evaluate or adjudicate Title IX-related concerns, and any AI-related harassment must be reported according to university policy using the same process as other forms of harassment.

Please note: Some departments and schools provide specific requirements or guidelines, such as these from the Medical School.

Considerations for Faculty

Given the sources outlined above, consider the following objectives when deciding how to address GAI with your students and how you use - or do not use - it in your courses.

Protect Student Privacy

You can use your GW login to access GW-supported AI tools and encourage your students to do the same. These versions better protect your data and are not used to train larger AI models.
Designing AI-related activities in ways that safeguard student information and privacy is best practice. Structure assignments so students are not required or encouraged to enter PII into GAI.
Configure all AI tools for privacy in their settings to opt out of saving data.
Consider informing your students how they should protect your intellectual property (IP) (e.g, whether they can add your course materials or their personal recording, ie: video or audio, of the class can be uploaded to an AI tool.)
If PII must be entered into an AI tool, use only GW-approved AI tools.
Consider offering an equivalent non-AI option for any assignment that requires the use of AI beyond those on the GW-supported AI tool list (and may pose greater privacy risks).

Ensure Accessibility and Inclusion

Encourage students to use GW-supported AI tools, which are free for GW students and have been reviewed for accessibility.
When integrating non-GW AI tools, prioritize options that are free and broadly available.
Resources to help you evaluate accessibility are available from the Web Content Accessibility Guidelines and the National Education Association.
Classroom policies addressing the use of AI should not infringe on approved DSS-approved accommodations and should allow for the use of assistive technologies that incorporate AI. Although if a student’s use of an AI tool fundamentally alters essential elements of the course learning objectives, contact DSS for a review and/or alternatives.

Prevent Discrimination

Recognize that AI tools reflect social and cultural biases, and address problematic or discriminatory AI-generated content when it arises.
According to GW’s Office of Access and Opportunity, biases from AI that are submitted in papers or other situations – and reported – are treated the same as any other bias report.
Respond to harassment or discriminatory behavior involving AI as you would any other student code of conduct violations through Conflict Education and Student Accountability (CESA).
Recognize that AI detectors have high error rates, produce false positives, and show potential bias against non-native English speakers. AI detectors should not be used alone as evidence for disciplinary action.
It is best practice to avoid using “AI traps” or deceptive detection techniques (e.g., hidden text or misleading prompts), which undermine transparency, may produce unreliable evidence, and can create inequitable or unfair disciplinary outcomes.

Rely on Human Judgement/ Retain Academic Authority

AI is best as a support for teaching, rather than a substitute for professional responsibility.
Faculty are fully responsible for course design and all decisions related to grading, feedback, academic standing, and access.
It is not recommended to allow AI to conduct investigations, adjudications, or determinations relating to misconduct.

Maintain Academic Integrity

It is best practice to communicate openly with students about the benefits and limitations of GAI.
Expectations for GAI use can be communicated to students in the syllabi and assignment instructions, including what forms of use are permitted or prohibited.
Consider modeling GAI transparency for your students by telling them when and how you use GAI in your course and ask them to do the same. Course directors and faculty may request in advance that AI users save a log of relevant prompts and output so that they can provide it in the event of further review of AI use.
Address violations of academic integrity involving GAI in a similar manner as other academic integrity concerns. This is typically done by submitting a Charge of Academic Dishonesty through the Conflict Education & Student Accountability (CESA).

Notes:

For more information on incorporating GAI in your teaching, see: Teaching and Generative AI.
As technology evolves, these policies will be updated. Contact teaching@gwu.edu if you have an update to be incorporated.