Assignments
Generative Artificial Intelligence and Assignments
Once you've chosen the policy framework for students' use of generative artificial intelligence (GAI) in your course, you will need to extend that guidance to your assignments and assessments.
(Framework inspired by Forbes, M. & Brandauer J. "What’s my stance on genAI in this class?" Gettysburg College Johnson Center for Teaching and Learning. Retrieved 8/20/2023 from https://genai.sites.gettysburg.edu/positions-and-policies)
Closed
If you choose to "close" a particular assignment to the use of GAI, you may want to articulate your rationale. Closing the assignment may mean that you will need to redesign it. You might consider using guidelines for the assignment that ask students to do what text-generating Large Language Model (LLM) tools such as ChatGPT, Bard, or Claude do not do well. For example, you could:
- Make the assignment very current, as some LLMs have knowledge cutoff dates of more than a year ago.
- Incorporate hyperlocal context, which may have intrinsic appeal to students and may encourage critical understanding about the LLM's capacity to yield worthwhile outputs.
- Add clear source and citation requirements if they differ from general course expectations.
- Add specific elements to a rubric that assess critical thinking.
You might also follow the guidelines of the Universal Design for Learning and focus on multiple means of expression. Students could:
- Create video or audio responses to the assignment, rather than text
- Display metacognitive skills by reflecting on their learning;
- Annotate in a text using Perusall or Hypothesis, which are integrated with Canvas.
While in-class written work and high-stakes assessments can effectively "close" assessments to GAI, keep in mind that these practices may put students with accommodations through AccessibleNU at a disadvantage and can exacerbate student stress.
Conditional or Open: Permitting or Requiring GAI Use
If your course is open or if you choose conditional use of GAI for your course, consider the following elements when permitting or requiring students to use GAI on assignments:
- Students should only be required to use what is available and free
- Share your reasoning for the use of GAI in the assignment and its value to students
- Ask students to use only data that are not private or personal (See Syllabus statement for instructors who engage students in using generative AI systems/software)
- Add a warm-up exercise to familiarize students with the tools you have chosen
- Be specific about how you would like the students to use the GAI and explain why: is it a starting point or idea generator? A debate partner? An editor for student-authored work?
- Incorporate reflective opportunities to inspire metacognition
- Consider the principles of the Universal Design for Learning and allow multiple means of expression (presentations, video essays, etc.)
- Familiarize your students with the Northwestern University Library citation guidelines for GAI
Some ideas for having students use LLMs in an assignment include:
- Start with an introductory exercise that provides an ethical use case of ChatGPT
- Brainstorm by asking an LLM questions about the material or subject (theories, frameworks, problems, etc.)
- Ask an LLM to brainstorm ideas for a project, such as a new business idea
- Ask an LLM to draft a children's story in a foreign language, then revise the prompt to create a story that uses advanced grammar
- Ask an LLM for feedback on their work
- Ask an LLM to summarize a source they have consulted
- Ask an LLM to analyze data they have gathered
- Ask an LLM to debug computer code
- Ask an LLM to write the code to clean large data sets, readying them for statistical analysis
- Prompt the LLM to take one side of a debate
- Prompt the LLM to write a sonnet on a particular topic and compare it with an existing sonnet
- Prompt the LLM to summarize a historical event, person, or period and have students discuss, correct, interrogate for accuracy and credibility
As part of the assignment, students could be asked to explain their use of GAI:
- Include a reflective paragraph on their LLM usage that details how they used it, what it provided and why it was or was not beneficial to their final product
- Include a copy of all prompts and text from the LLM as an appendix
- Identify issues of bias, relevance, and accuracy that they encounter while using an LLM
- Post using Discussions in Canvas to share work with the LLM while it is in progress
Example: Ask ChatGPT to write an essay on your topic
"You have been researching a particular topic for your final presentation. I'd like you to ask ChatGPT 3.5 to write a 500-word essay on the historical importance of your topic. Copy that essay into Word, along with the prompt you gave it. Then turn on Track Changes in Word and edit the essay: correct any errors, verify any facts that ChatGPT cited by putting a comment on the fact and showing at least one other source for it, improve the writing to make the essay clearer and more interesting. Then write a paragraph or two titled "Feedback" that explains your overall assessment of the ChatGPT essay and give it a grade. Submit your Word document to Canvas."
Rationale: This assignment falls part way through a course, after students have developed expertise on a topic. By engaging with ChatGPT, students will:
- Express their expertise by corroborating or debunking the items in the essay
- Use historical analytical skills by verifying facts and checking other sources
- Use editing skills to improve writing
- Put themselves in an evaluative mode and explain their thinking.
Consider the following examples of assignments that have been adapted to make use of GAI in ways that will advance learning:
Example 1: A communication and marketing plan
Part of the original assignment asked students to develop a communication and marketing plan, which took about three weeks.
The revised assignment instructed students to ask ChatGPT to draft multiple communication and marketing plans. Next, the students are required to analyze the results; identify, with justification, the best elements of the various plans; and adapt these into a single plan.
Rationale: This assignment may shorten the amount of time devoted to the nuts and bolts of the assignment - developing the plan - allowing the students more time to evaluate, analyze, and synthesize.
Example 2: A lab report
Part of the original assignment asked students to gather data and write a lab report detailing the purpose, methods, and findings of their experiment.
In the revised assignment, students were given an editable, ChatGPT-generated lab report as an example of "C" quality work. In addition, they were familiarized with the rubric for evaluation. Students were asked to update the lab report with their own results, edit its analysis, and try to improve it from "C" to "A" quality work.
Rationale: This assignment will not shorten the amount of time devoted to the laboratory work, but it may deepen students' analysis and editing skills.
Video Examples
- Ignacio Cruz, Assistant Professor, Communication Studies, "Classroom Activity: AI-Enabled Hiring."
- Ken Alder, Professor, History, "Assignment: Using ChatGPT for Research Projects."
See the Northwestern University Writing Program AI Resources site for an extensive list of ways to incorporate generative AI into writing assignments.
Back to top