Artificial Intelligence (AI)
AI-based, Large Language Models (LLM) such as Google’s LaMDA, and OpenAI’s ChatGPT will continue to progress and be integrated with many computer programs that we use every day. These advancements bring productivity benefits as well as pedagogical challenges. Instructors at UMD are encouraged to continually adapt to and embrace the use of advanced technologies. AI has the potential to contribute to teaching and learning in meaningful ways, and in other instances be a deterrent when skill development requires that its use be restricted.
At TLTC, we look forward to helping you think creatively about your assessments and your specific learning outcomes to put authentic, relevant, student-centered learning at the forefront of your academic planning. Here are important points to consider as you plan your course:
- Speak openly and frequently with your students about your expectations for technology use — specifically, for AI-based tools such as Large Language Models.
- Like many other software/apps available on the web (free or not), access to AI-based tools requires users to agree with specific terms of service and privacy policies. Instructors and students should carefully read these documents and understand the risks associated with their use, prior to accepting the terms.
- If your pedagogy allows you to embrace the use of AI in your course, avoid making its use a requirement, unless you plan to also offer alternative ways to achieve similar outcomes.
- Integrating AI tools, or motivating compliance with relevant rationale, will be more effective than an adversarial approach to restrict and detect AI-generated content with technology solutions.
- Turnitin - UMD’s plagiarism tracking software, is working to enhance its capacity to detect AI-generated content, should you need to deter its use in your course.
- Want help with your course? Request a consultation with TLTC.
- Concerns regarding student work? Refer an issue to the Office of Student Conduct.
Considerations for Teaching
Test your assignments and adjust as needed
Consider putting your prompts into ChatGPT to view the results. If you have concerns, you might try the following::
- Scaffold assignments: Ask for an outline, first draft, as well as final paper. Structure assignments to allow for peer review.
- Create nuanced or authentic assessments: Incorporate current events or personal experiences. Convert a written assignment into an infographic, video presentation, podcast, or drawing.
Consider incorporating AI
AI-based tools such as Large Language Models (LLM) may be leveraged in big or small ways. Some example assignments include:
- Allow students to use a LLM to create a draft, then ask them to deconstruct and edit the draft and turn in both versions.
- Ask students to submit an outline and create AI generated drafts for them.
- Encourage students to use LLM to search terms when beginning their research process.
- Have students generate an argument on a topic using LLM, then write a counterargument.
Decrease incentives to cheat
Use teaching practices that increase motivation through high expectancy-value:
- Set realistic goals (Parsons et al. 1980:. Make sure your students have enough time and support to thoughtfully complete their assignments.
- Describe the relevance of the skills (Hullerman et al., 2016): Help students understand what they gain from various parts of your assignments
- Offer options to increase choice and control (Patall et al., 2010): When applicable to your course goals, allow students to choose their topics or modes of delivery for an assignment (e.g. an essay, presentation, short video, drawing, or infographic)
- Consider alternative grading strategies: Many alternative grading strategies are designed to help students focus more on their own learning and the course content rather than grades.
Decide how you want your students to engage or not engage with AI and communicate these expectations clearly and frequently.
- Be consistent: Make sure you are conveying the same message in writing on your syllabus and assignments as well as verbally in class.
- Explain your reasoning: Share both what your policy is and why you have made those decisions. If you are prohibiting them, about the skills students miss out on when they rely on AI-based tools. Discuss why you might have different expectations for using AI-based tools in different assignments or in different capacities.
- Discuss misconceptions: Directly point out any common pitfalls or misconceptions around the use of AI chatbots in your field. Discuss websites tools such as ChatGPT, QuillBot, or DALL-E directly so students know exactly how they work.
See the additional resources section further down on this page for suggested syllabus language.
Turnitin software or other non-enterprise tools like Hugging Face and GPTZero are currently not effective at detecting AI text if it is modified by students. Instead, you might look for changes such as:
- Dramatic change in tone or style
- Code has excessive number of bugs
- Technically correct but surface-level writing
If your students know you are fully engaging with their work and paying attention to the assignments they turn in, they will be less likely to engage in academic dishonesty.
If you have concerns regarding student work, refer the issue to the Office of Student Conduct. They will work with both you and the student to learn more, potentially investigate the incident, and guide the student toward more appropriate academic conduct.
Below you will find a recording and resources from a January 2023 workshop.
Depending on how you may or may not want your students using AI-based tools such as ChatGPT, here is some sample language:
Option 1 (no AI): In this course, my expectation is that you will not use any artificial intelligence (AI)-powered programs such as ChatGPT or DALL-E to help you with your assignments. Any use of AI-generated work to outline, write, create, or edit your assignments will be considered an academic integrity violation. My reasoning for this is that these programs may provide inaccurate or biased information, but more importantly, they do not serve your development as a student. In this course you will learn valuable skills from outlining, generating, and editing your own work. If you have any questions about this policy or are not sure if a resource you have found will violate this policy, please ask.
Option 2 (some AI): In this course, I encourage you to use artificial intelligence (AI)-powered programs such as ChatGPT or DALL-E to help you with some assignments. When you use these tools, it is your responsibility as a scholar to make sure you are clearly communicating the AI involvement in your work. Please make sure to use phrases such as “[your name] via DALL-E 2” (for images) or “This paper was generated with the help of GPT-3” (for essays). Please review the instructions in each assignment for more details on specifically how to show your work.
- UMD's Undergraduate Studies' Faculty Readiness Rubric for Academic Integrity
- UMD's Office of Faculty Affairs’ guidance on research integrity.
- How to Productively Address AI-Generated Text in Your Classroom (Indiana University Bloomington)
- Artificial Intelligence Tools and Teaching (University of Iowa)
- ChatGPT and AI Composition Tools (Washington University in St. Louis)
- Why I'm Not Scared of ChatGPT (The Chronicle of Higher Education)
- Brown, S. (2021, April 21). Machine Learning, explained. MIT Sloan. Retrieved January 4, 2023, from https://mitsloan.mit.edu/ideas-made-to-matter/machine-learning-explained
- Crawford, K. (2021). The atlas of AI: Power, politics, and the planetary costs of artificial intelligence. Yale University Press.
- Heikkilä, M. (2022, December 19). How to spot AI-generated text. MIT Technology Review. Retrieved January 4, 2023, from https://www.technologyreview.com/2022/12/19/1065596/how-to-spot-ai-generated-text
- How should I credit DALL·E in my work? OpenAI Help Center. (n.d.). Retrieved January 4, 2023, from https://help.openai.com/en/articles/6640875-how-should-i-credit-dall-e-in-my-work
- Hulleman, C. S., Barron, K. E., Kosovich, J. J., & Lazowski, R. A. (2016). Student motivation: Current theories, constructs, and interventions within an expectancy-value framework. In A. A. Lipnevich et al. (Eds.), Psychosocial Skills and School Systems in the 21st Century. Switzerland: Springer International Publishing.
- OpenAI. (2022, December 21). ChatGPT: Optimizing language models for dialogue. OpenAI. Retrieved January 4, 2023, from https://openai.com/blog/chatgpt/
- Parsons, J. S., Futterman, R., Goff, S. B., Kaczala, C. M., Meece, J. L., & Midgley, C. (1980). Self-perceptions, task perceptions and academic choice: origins and change (pp. 198). Ann Arbor, MI.
- Patall, E. A., Cooper, H., & Wynn, S. R. (2010). The effectiveness and relative importance of choice in the classroom. Journal of educational psychology, 102(4), 896.
Page last updated February 2, 2023