Considering Generative AI and ChatGPT at Virginia Tech

You may be hearing an array of opinions about the implications that generative AI (Gen AI) tools might have for the future of teaching and learning. Any time a new technology like Gen AI emerges and captures the public’s attention, initial reactions tend to be polarized. Some see AI as a threat to higher education, while others view it as a positive force for reshaping the way we teach.

In the midst of these competing and conflicting perspectives, TLOS recommends a measured approach. We ask faculty to consider the following suggestions as you make decisions about the courses you teach. Because these tools are changing rapidly, we will continually evaluate their use at Virginia Tech and revise our guidance as the situation evolves.

What is generative AI?

Generative artificial intelligence (Gen AI) describes several types of AI that are capable of creating new text, images, code, audio, or other types of content. Tools such as ChatGPT have been trained using very large collections of existing media to generate human-like responses to user-authored prompts. For text-based content, these large language models can be used to answer questions, write essays or articles, create lesson plans, and much more. The content creation process is limited by the tool’s training media and the user prompt.

Why does it matter for Virginia Tech?

Gen AI tools are becoming increasingly influential across many industries, making it essential for students to learn how to utilize these tools within their specific fields of study. However, the potential for misuse of Gen AI also raises concerns for evaluating student learning. Interrogating the risks and opportunities of these tools in different types of courses and disciplines (course size, delivery modality, topic, etc.) is key to finding ways to leverage their power while limiting their abuse. 

The true impact of these technologies remains to be seen. It is not likely to cause the end of higher education, and it is not likely to be the panacea some expect it to be. Deciding how to apply and constrain these tools is likely to be a nuanced issue that will require thoughtful and informed debate.

Recommendations

1. Become familiar with generative AI tools in order to have an informed perspective.

• See the links below for some basic introductions to the technology and its implications.
• Attend a workshop or forum discussion on the topic.
• Reach out to support staff in TLOS, CETL, the Office of Undergraduate Academic Integrity, and the University Libraries for specific advice.

2. Consider the Honor Code and its applicability to generative AI tools.

• The Office of Undergraduate Academic Integrity offers the following guidance: While most students largely engage in honest behavior in the classroom, some may choose to use tools such as ChatGPT to engage in academic dishonesty. Please continue to be clear in your expectations with your student related to the Undergraduate Honor Code and the use of AI software just as you would other websites that may provide students with means to engage in academic dishonesty. The unauthorized use of ChatGPT and other AI software may fall under several definitions of academic dishonesty in the Undergraduate Honor Code. If you believe that a student has engaged in academic dishonesty, please contact the Office of Undergraduate Academic Integrity at 540-231-9876 or email the office at honorsys@vt.edu to discuss what occurred.
• The Graduate Honor System reviews each case in its own context. Faculty are encouraged to provide clear and precise guidance about when or if the use of tools such as ChatGPT are allowed, encouraged, or prohibited and candidly discuss with students the learning value of completing an assignment with or without such tools. GHS decisions about any potential violation will be based on what guidance faculty provided. If you believe that a graduate student may have committed an academic integrity violation, please refer the case to the Graduate Honor System; 540-231-9564; ghs@vt.edu.

3. Avoid being drawn into a confrontational mindset regarding these tools.

• Given the obvious implications for academic integrity, tools are being developed and promoted that may provide some perspective on whether or not student submissions were written by AI. Due to the evolving nature of AI solutions, these tools are never going to be 100% accurate, and an overreliance on a tool to “catch” anyone using AI is likely to lead to a slippery slope of ineffective and highly stressful interactions with students.

4. Set clear expectations for your students regarding the use of generative AI/ChatGPT.

• Directly address the reasoning behind any restrictions you decide to implement.
• Consider updating course documentation about academic integrity to include a statement specifically addressing the use of generative AI/ChatGPT (see ideas below).

5. Explore potential changes to your course design and/or assessment strategies.

• Consider incorporating course assessment practices aligned with principles of authentic assessment such as seeking to replicate professional experiences, expecting a task-focused demonstration of understanding, and providing opportunities for practice with feedback. (See Teaching Resources - Assessing Student Learning - Authentic Assessment from Indiana University for some more detailed ideas.)
• Consider allowing multiple options for action and expression in your assessments. This is a principle of Universal Design for Learning that suggests students be allowed to select from various methods and technology tools to showcase their understanding and support their learning.
• Proactively incorporate generative AI/ChatGPT into your instructional strategy, inviting students to use these tools to spark creativity, or to test and evaluate the accuracy of their output.

In the end, our suggestion is that these tools represent an evolution of existing information technologies and that we should carefully consider how to utilize them to improve our shared experience.

Additional Resources on Generative AI

ChatGPT Overviews for Faculty

• ChatGPT and other AI-based tools - Center for Teaching and Learning (University of Washington)
• Using ChatGPT in Education: Guide for Teachers (Edtech Empire Education Technology Blog)
• ChatGPT: Understanding the new landscape and short-term solutions (Cynthia Alby)
• Teaching Students to Write with AI: The SPACE Framework (Medium.com)
• ChatGPT Advice Academics Can Use Now (Inside Higher Ed)
• ChatGPT Resources for Faculty – University Center for Teaching and Learning (University of Pittsburgh)
• The Mechanical Professor (Ethan Mollick)
• Guidelines for Using ChatGPT and other Generative AI tools (Harvard Office of the Provost)
• Generative AI Resources (University of Michigan)
  • Guidance for Faculty/Instructors (University of Michigan)
• AI: Considerations for Teaching and Learning (The Ohio State University)

Virginia Tech Honor Code

• Definitions of academic misconduct

Virginia Tech Graduate Honor System

• Graduate Honor System Constitution

More Information on Authentic Assessment

• Teaching Resources - Assessing Student Learning - Authentic Assessment (Indiana University)
• Authentic Assessment Toolbox Home Page (John Mueller)

Potential starting point for syllabus statements (fittingly, generated by ChatGPT):

"While the use of AI tools, including ChatGPT and other AI-powered writing or research assistive technologies, may be tempting, it is important to consider the ethical implications of their use. In this course, students are encouraged to consult with the instructor before using any such tools. If approved, the use of AI tools must be properly cited and acknowledged in all submissions. Failure to follow these guidelines may result in a grade penalty or other disciplinary action as determined by the instructor."