Designing Assignments and Activities with ChatGPT and Generative AI in Mind

Generative AI, such as ChatGPT, can be a powerful tool to engage students in learning and creativity. Essentially, generative AI tools are those that create content on their own without human intervention. It can be useful for writing text, generating ideas, creating images, writing and editing code, and more. By designing assignments that incorporate generative AI technology, instructors can provide students with opportunities to explore, create, and problem-solve. However, as an instructor, you may also want to create assignments that challenge students to demonstrate their own knowledge and skills without relying heavily on AI-generated content. In this article, we will review different assignment ideas and strategies to create prompts and assignment ideas in different disciplines.

Table of Contents

Workshops

Syllabus Statements and Student Input

Intelligent.com conducted a poll of more than 1,000 current college students in May 2023 regarding their use of ChatGPT for coursework. 30% of students used ChatGPT for coursework during the 2022/2023 academic year, and of that group, 46% utilized it frequently. Users of ChatGPT and other generative AI tools like Bing Chat and Google Bard continue to grow with some flattening of the upward trend in summer 2023. Generative AI is rapidly advancing and becoming more prevalent in education, work, and our daily lives. As an educator, it’s a good idea to help students be aware of the ethical considerations surrounding the use of generative AI.

  • Consider adding an acceptable use statement in your syllabus. Here are some guidelines and examples.
  • Consider having a discussion with students at the beginning of your course and before major assignments to discuss the ethical use of AI tools. Possible discussion questions:
    • How do you think generative AI can be applied to the course assignments in this class?
    • Can you share any specific examples of generative AI being used in educational settings?
    • How can we ensure that AI tools are used in a way that promotes skill development in our course?
    • After reviewing the assignment directions and grading information, what would be some helpful uses of AI tools that will still allow you to learn the content and demonstrate your learning?
  • Based on various surveys and instructor experiences, not all students believe it is ethical to use AI on assignments. Be sure to include a discussion/policy about how AI can or cannot be used in group work.

Is AI Use Cheating?

There is no standard for determining if AI use by students qualifies as plagiarism or cheating. There is also no consistent standard for citing or crediting work using an AI tool. It may be useful to check with your professional organizations and journals and share any of their policies with students. Currently, AI is part of retail and other business careers, education in personalized learning, systems that make recommendations, human resources decisions, healthcare, agriculture, gaming, marketing, finance, and more.

Organization and publication examples:

Citation Style Guidance:

It may be useful to reflect on how you define plagiarism and cheating and then help guide students to think about it. Review this image from Matt Miller @DitchThatTextbook to help guide your thinking.

Plagiarism and cheating graphic with a spectrum showing "Bot-Created" to "Student-Created" to help guide teachers in thinking about what counts as plagiarism and what does not.

No True Detection of AI is Possible

There is no “fool-proof” way to detect AI use in student projects, and there have been many stories published about false positives and negatives using various AI detectors.

At NC State University, we provide access to Turnitin, which has an AI detector if you would like to get some input on if students have used AI to craft their writing. That said, do not use Turnitin as sole evidence that a student has cheated or plagiarized. Please review the academic integrity guidance and policies from the Office of Student Conduct. Note that the Division of Academic and Student Affairs also encourages faculty to notify students if they plan on using Turnitin.

AI detection and AI detector workaround programs are regularly being created and released. Here are some common tools and videos guiding students and content creators on how to get around AI detection.

There are also some red flags you can look for in reviewing student work. It’s helpful (albeit difficult in large classes) if you know your students writing and can determine if an assignment does not fit their typical way or level of writing. What to look for:

  • A factual error or made-up citation
  • Missing required assignment data sources or article text
  • “Too perfect” in terms of grammar and usage
  • Overly formal, detached, or impersonal style/tone
  • Predictable formations – -like a five-paragraph essay from middle school language arts
  • The writing too directly and repetitively parallels the assignment directions

Note: Students who are good at prompt writing and provide context, follow-up questions, a voice for the AI, etc., may not produce writing that exhibits these flaws. You may also want to consider having a conversation with a student about their work and topic if you have concerns. ChatGPT-4 (a paid option) is significantly better at avoiding these style issues, and Bing Chat is powered by GPT-4 (free).

Designing Assignments to Limit AI Usage

There are ways to design assignments that can make generative AI use more difficult for students. However, as tools become more sophisticated, assignment revisions may not be enough to truly prevent students from using AI; however, these strategies are a good start.

Ask ChatGPT

Ask ChatGPT to provide assignment examples in your field that would be difficult for it to complete. Include context, specific learning outcomes, and more to get a more specific list of suggestions. Prompt Example:

  • You are a professor for an introductory course in {subject area} at the college level. You are trying to design assignments that would be tricky for students to use AI to complete. What are some assignment ideas and topics within the field that would be difficult for Bard to complete successfully?
  • You are a professor for a college statistics course. Students are expected to recognize and be able to explain the central role of variability in the field of statistics. They also must be able to find variability when interpreting data. What are some course assignments that students can complete to show they have met these objectives and that are difficult for ChatGPT to complete? Explain how the assignment will help students demonstrate their understanding and what makes it complicated for a generative AI tool like ChatGPT. See the results here!

Require Specific Data Sources to be Used in the Assignment

ChatGPT is not connected to the web. It’s a “pretrained” tool that has not been trained on information post-2021. So, incorporating specific texts into assignments can make things more difficult for ChatGPT. You can ask students to write and cite sources/text from specific articles or videos. You can also provide data sets that students must use in their work.

Google Version History

Require that students submit written work using Google Docs, Slides, Sheets, etc., and use version history to validate that the writing and input occurred over time vs. in large chunks suggesting that students may have copied and pasted from another source like ChatGPT. Students have also used time stamps in Google Docs version history to exonerate themselves from false positives picked up by AI detectors.

Incorporate Student Discussion and Collaboration

In-person student discussions that reference past class activities, readings done outside of class, previous lectures, and so on can be integrated into your course. Examples:

  • Ask students in a chemistry course to compare and contrast two models that they read about for homework or that were shared in a recorded lecture. Ask students to come up with examples in class (or on a discussion board) with a partner based on the reading assignment.
  • Use Perusall and set the auto-grading (ai-assisted) feature to highly weight active engagement time and getting responses. Manually grade and let students know that credit comes from their in-text conversations with each other.

Reflective Assignments

AI tools are not truly reflective and aren’t likely (even fictionally) to make good connections between course content and personal experience or learnings. Examples:

  • Write a reflection on a time when you struggled with a {subject area} concept. What was the concept? How did you eventually understand it? What advice would you give to other students who are struggling with the same concept?
  • Compare and contrast two different ways of solving a problem {in your content area}. What are the advantages and disadvantages of each method? When would you use one method over the other?

Real-World & Localized Connections in Assignments

Some AI tools are not connected to the internet and will not have an understanding of local references or the most recent sources. Others may not be able to draw connections that make sense to humans who understand those “smaller” contexts. For example, we asked Bard to write a short story set in a modern-day context in Raleigh, North Carolina on the NC State Campus and gave it some specific guidelines. In addition to writing a formulaic story, Bard regularly referenced “The Old Well” which is part of the UNC-Chapel Hill campus. Prompt example:

  • Analyze the impact of a recent policy change {content-specific} or ask students to choose a policy change that has been implemented in the last year. Research the policy change and its implications for the economy. Write a report that includes the expected impact, strengths and weaknesses of the change, and recommendations for how the policy change could be improved.

Take Assignments through a Process

Asking students to complete an assignment with a process including steps like brainstorming, mapping, drafting, peer review, an interview, and a final product can make it difficult for them to find successful ways to use AI. It may be able to help students with sections of the assignment but not the entire product or process. You can also ask students process-oriented questions along the way. You can also include ambiguous questions or those that require positions on controversial topics. Examples:

  • Compare your answers to your team’s answers. Discuss any differences.
  • Explain the process you followed to arrive at your conclusion.
  • Analyze the ethical implications of each step in the process and propose alternatives if necessary.
  • Explain the long-term consequences of implementing this process and how they might evolve over time.
  • Discuss the role of creativity and innovation in…
  • Identify potential biases, assumptions, and problems that could arise and suggest methods to mitigate them.

Retrieval Practice Activities

Retrieval practice activities allow students to practice recalling information from class activities, lectures, readings, and so on. If specific to course content, AI would not be helpful in these activities (particularly if completed in person). More on retrieval practice.

Multi-Step with a Creative Component

Create projects in which students demonstrate their learning. Essentially find ways to ask them to take what they’ve learned, organize it, and make something with it. Examples:

  • Short story writing in which students must use content information, specific vocabulary, and maybe even primary sources to craft a story.
  • Ask students to create a comic strip based on a concept, vocabulary, a reading, etc.
  • Students creating a public service announcement video to demonstrate learning

Blended Instruction or Flipping

You may also want to consider using blended or flipping formats for your course in order to limit AI use. In this model, students would learn content outside of class time and then use class time for the application of what they learned.

Outside of Class ActivitiesIn-Class Activities
Interactive lectures and readings
Quizzes for learning
Practice activities
Preparation for class discussions
Preparation for class presentations		Class discussions, Think-Pair-Share
Retrieval activities
Group project work
Presentations
Case study work
Reflection
Assessment

Designing Assignments to Work with AI

AI tools are likely to be used by students in future careers and likely in their coursework, so one approach is to incorporate the tools directly and intentionally into assignments and activities.

“Am I going to teach students to write or to write with AI tools like ChatGPT?

Derek Bruff

Consider these assignment reflection questions from Derek Bruff’s article “Assignment Makeovers in the AI Age.”

  1. Why does this assignment make sense for this course?
  2. What are the specific learning objectives for this assignment?
  3. How might students use AI tools while working on this assignment?
  4. How might AI undercut the goals of this assignment? How could you mitigate this?
  5. How might AI enhance the assignment? Where would students need help figuring that out?
  6. Focus on the process. How could you make the assignment more meaningful for students or support them more in the work?

Consider these ideas for assignments that can work with AI tools:

  • Use AI to generate multiple explanations for a concept and ask students to critique the AI-generated explanations. Ask them to cite/use specific course readings, notes from lectures, etc., in their critiques.
  • Save time in reviewing student writing by asking them or requiring them first to get an AI review of their work, then reflect on the review, make edits, and then submit their final work.
  • Include an AI tool in a “Think-Pair-Share” activity in class. Students pair with another person in class and then with an AI tool.
  • Ask students to predict what responses they will get from AI to specific course content questions, problem sets, etc.
  • Provide several responses from AI and ask students to make a better or different product using those drafts/responses. They might make a mind map from a narrative created by AI and then find three additional sources to support or expand on different sections of the mind map.
  • Assign a peer teaching project in which students will teach a concept or review a concept for their peers. Encourage students to get help from AI with the content and in designing a short activity that can be done as part of the peer teaching. Make students responsible for answering questions from peers and instructors. Use any gaps to adjust your own teaching.
  • Ask students to debate an AI tool — students on one side and ChatGPT on the other.
  • Ask students to find evidence for an AI-created “main points” of an article. First, copy and paste an article into ChatGPT (or a link to an article into Bing or Bard) and ask the tool to summarize the key points of the article. Then provide that to students and ask them to find quotes or details that expand on each point.

Resources

NC State Office of Faculty Excellence: Navigating the Landscape of Higher Education in the Age of Artificial Intelligence

Writing Instructors –> Tim Laquintano, Carly Schnitzler, and Annette Vee — TextGenEd: An Introduction to Teaching With Text Generation Technologies (Assignment examples for AI Literacy, Creative Explorations, Ethical Considerations, and more – access at the bottom of the article)

Writing Instructors –> Anna Mills (Curator). AI Text Generators and Teaching Writing: Starting Points For Inquiry

AI Writing Detection: Red Flags

Ethan & Lilach Mollick — Using AI to Implement Effective Teaching Strategies in Classrooms: Five Strategies, Including Prompts

Ethan Mollick — Assigning AI: Seven Ways of Using AI in Class and The Homework Apocalypse 

Jeffrey Young — EdSurge Instructors Rush to Do ‘Assignment Makeovers’ to Respond to ChatGPT” 

Derek Bruff

Tyler Cowen & Alexander Tabarook How to Learn & Teach Economics with Large Language Models, Including GPT

Sam Lau & Philip Guo Teaching Programming in the Age of ChatGPT – O’Reilly 

AI Prompts for Teaching 

Impact Research: K-12 Teachers & Students ChatGPT Use

Torrey Trust — Essential Considerations for Addressing the Possibility of AI-Driven Cheating, Part 1 | Faculty Focus 

Ideas to Limit AI Use in Assignments from Google Bard 

Educause Review: Artificial Intelligence

An introduction to prompting generative AI like ChatGPT for teaching and learning 

ChatGPT, Chatbots and Artificial Intelligence in Education – Ditch That Textbook

Artificial Intelligence and the Future of Teaching and Learning (PDF) 

Rethinking your Problem Sets in the World of Generative AI – MIT

Hybrid Teaching: Best Practices

Blended Learning | Columbia CTL 

How We Use AI to Enhance Your Writing | Grammarly 

30 AI tools for the classroom – Ditch That Textbook 

College of Education ChatGPT Resources

Back to top