Guidance on the Use of AI in Teaching and Learning

The goal of this document is to provide reasonably thorough, but also accessible and brief, guidance to support the AI approaches and practices of instructors, as well as how they communicate their expectations to students. The guidelines represent evidence-based practices that align with Purdue’s values and policies. We encourage students and faculty together to innovate learning through AI, and we are eager to work with and support student innovation in AI.

Thank you to all who provided feedback on the Spring 2024 guidance via Qualtrics. The guidance for instructors will be updated each semester and is expected to evolve as we gain experience, advance the technology and implement innovations. In addition, based on your input we will develop equivalent guidance specifically for students.

Our Purdue community of faculty, staff, and students are invited to a session hosted by the Teaching & Learning Community of Practice (TLCoP) for further discussion on the AI Guidance. See the TLCoP website for details and registration.

We would also like to thank the individuals who provided input, including members of Purdue's University Senate, the Center for Instructional Excellence, and the Innovation Hub.

AI in Teaching and Learning Resources

Additional resources for Purdue instructors can be found in the Artificial Intelligence in Education module of Teaching@Purdue. If you have ideas for new resources and content, please email InnovativeLearningTeam@purdue.edu.

To receive monthly updates, sign up for the AI Teaching & Learning Digest letter . The digest will contain national and regional news in Teaching & Learning with AI including updates on emerging technology, curated journal and pre-print research articles, and stories and experiences of Purdue faculty and staff who are experimenting with AI in their courses.

Existing Relevant Policies and Regulations

For convenience, below are links to existing university policies and regulations that are re levant to the use of AI in Purdue learning environments:

• Libraries Policies on Acceptable Use of Electronic Resources
• Intellectual Property (1.A.1)
• Use of Copyrighted Materials for Educational and Research Purposes (I.A.3)
• Acceptable Use of IT Resources and Information Assets (VII.A.4)
• Academic Regulations on Grades and Grade Reports
• Student Conduct

Syllabus Language

Now is a time of rapid innovation and change. Good instructor-student communication is at the heart of all great teaching. Instructors should strive for maximum transparency and clarity in their communications to students about the use of AI in their course.

Purdue’s student conduct regulations state the following regarding academic integrity:

• “Students are expected to adhere to the guidelines provided by instructors for academic work so that no student gains an unfair advantage. Using or attempting to use unauthorized materials, information, study aids, notes, or any other device in any academic exercise will not be tolerated. Unauthorized materials may include anything which or anyone who gives a student assistance that has not been approved by the instructor in advance.”

The instructor, therefore, should determine and inform students when the use of AI is authorized or unauthorized. Authorization is determined at the course level by the instructor.

On the course syllabus, instructors should include specific guidance on:

• Allowable use of AI tools by students for assessed work in the course. Example: use of AI while completing a homework assignment.
• Appropriate use of AI tools by students during the learning process for non-assessed work of the course. Example: use of AI to summarize an assigned reading.
• The ways in which the instructional team may attempt to detect the use of AI on assessed and non-assessed work. For example: Non-assessed work could include the early assignments in a scaffolded project with several components. (see below)
• The potential consequences faced by students who violate the AI use policies in the course.

See “AI and your Course Syllabus” on Teaching@Purdue for sample syllabus statements . If you have new syllabus examples you feel may be valuable to share with others , please send them to InnovativeLearningTeam@purdue.edu.

Instructional Team Roles and Responsibilities, Grading

The use of AI tools may change the workflow for the instructional team by shifting the kind of work they do, as well as who might be accountable for completing certain kinds of work. Typical principles of teaching and learning still apply:

• The instructor of record is responsible and accountable for all grades on all assessments in the course.
• All assigned grades must be explainable and defendable based upon evidence submitted by the student for the assessment.
• Some grading responsibilities may be delegated by the instructor of record to other members of the instructional team (ex: teaching assistants) based upon their professional judgment, course enrollment, and/or the complexity of the assessment. The instructor of record retains the responsibility for oversight of other graders, training them appropriately, and ensuring their work is both consistent and of a high standard.

In the presence of AI tools used in grading student work, the instructor of record and their instructional teams should:

• Create and apply a process by which any AI-based tools used for grading student work produce explainable and defendable grades. This process should:
  • Ensure that a human remains in the loop to the maximum extent possible.
  • Integrate a quality assurance procedure to ensure grading automation features (ex: Gradescope’s grouping tool) work accurately and as intended.
  • Avoid “blackbox” grading methods/tools that cannot be validated or explained.
• Consider the appropriateness of using AI detection tools (see below).

Use of AI Detection

Current AI detection tools have very high false-positive rates, rendering them almost useless in practice. At the present, instructors should treat any results generated via AI detection tools, including those in TurnItIn, with extreme distrust and carefully review suspected cases with due diligence and according to a procedure clearly explained to students on the syllabus.

On the course syllabus, instructors should:

• Clearly describe the specific ways in which they will (or will not) attempt to detect the unauthorized use of AI.
• The procedures they will follow if an unauthorized use of AI is suspected. It is especially important, given the high false positive rates of current AI detection tools, that a positive result from one of these tools is not the sole determining factor of an academic integrity violation and subsequent consequences.
• The potential consequences faced by students who are determined to have used AI tools in an unauthorized way, including any impacts on grades and possible referral to the Office of Student Rights and Responsibilities.

Copyright Issues

Some AI based tools encourage users to upload copyrighted material as training data for specific AI models, particularly those used to create content for ‘personalized learning.’

On the course syllabus, instructors should:

• Make clear to students that sharing of copyrighted material with third-party AI tools is prohibited.
• Remind students that while faculty and instructors do not own copyright to facts or ideas in their discipline, they do own copyright to their expression, explanation, and presentation of those facts and ideas in course notes, PowerPoint slides, etc. including assessments constructed for the course. As such, those instructor-generated materials should never be uploaded to any third-party site (whether AI oriented or not).

FERPA and Other Privacy Issues

Some AI based tools, including tools that promise AI-detection capabilities, use user-supplied data to further train their models. Third-party AI tools have widely-varying approaches to privacy, security, encryption, and so forth, and instructors must–as always–comply with FERPA requirements for student information and privacy. 

On the syllabus, instructors should:

• Clearly communicate with students how their data or interactions with AI tools may be used or shared before an instructor puts any of the student’s work into an AI tool, including AI detectors
• Commit to never sharing personally identifiable information about students with any third-party AI tool
• Explain that they will obtain students’ consent to share data on their interactions with AI when appropriate. Example: a research study on student use of AI tools

The university is working toward expanding our technology-based teaching and learning environments where instructors and students can confidently work with copyrighted instructional materials and FERPA-covered student content to build, deploy, and use AI tools for educational purposes. Please plan to bring your ideas and teaching technology environment needs to the Spring 2024 town hall.