Using Generative AI in Student Assignment
Instructors can ask students to use generative-AI tools for assignments in intentional ways that support their learning development. (The link provided demonstrates how ChatGPT was used to ask about using these tools to support student learning.) Some ways in which the tools are proving particularly useful for students:
- Generating an outline for an assigned topic. Students work in groups to workshop the quality of the outputs.
- Support for writing draft essays: students can get feedback on their drafts. Student groups can workshop the generated responses to develop skills in critical thinking and improve assessment of writing.
- English language learners can ask for review of spelling and grammar and idiomatic usage.
- Support for understanding difficult texts: one of the things ChatGPT and other text-based models have proven generally good at is text summary. Students can input text that they’re struggling with and request a summary.
The ethics of using generative AI for creating original work are complicated, from the question of whether what these tools create should be considered plagiarism, to the ways the tools mirror biases and stereotypes in Western society, to who “owns” the work created, to the ethics of how the tools use works created by others. It’s important to raise these questions in open conversation with your students and, as faculty, academic researchers and creative artists, with each other. As yet, there are no real answers, but exploring the tools through guided activities will help students understand the uses and pitfalls of generative AI tools along with developing a sense of the larger cultural and ethical questions.
Some things you may want to do to help students understand both the uses and problems of generative AI tools:
- Have student groups workshop AI responses and critique their quality
- Have students reflect on what a particular assignment is intended to achieve and consider how these tools would help or hinder achieving those outcomes
- Help students learn to create effective prompts.
Below are examples of some ways faculty have incorporated AI tools into learning activities. These activities are about helping students use AI as a tool for learning, not about teaching students to use AI in general. If you want to try using AI tools in your class, try prompting a text-based AI tool like ChatGPT to create an assignment. Faculty should practice crafting prompts in different ways to see how the results vary. In general, it’s advisable to begin the prompt by telling the tool “who” it is, e.g., “You’re a faculty member in a university teaching a course on X and you want to design an assignment that will lead to A, B, and C learning outcomes.” These are just a few of myriad possibilities drawn from faculty in the New School and elsewhere.
Courses requiring students to submit written assignments
Compare and contrast an essay on the same subject from an AI tool vs. a higher quality human written essay, and then provide feedback on how the AI essay could be improved. Have students use AI to write a first draft and then edit it, using track changes and comments to show their process OR ask students to write a first draft and have AI edit it.
Courses requiring students to analyze images
Compare and contrast AI-generated images prompted to create in the style of a specific artist, with actual images by those artists. Discuss in small groups then report back: Describe the “aesthetic choices” generated by AI when it is prompted to make art in the style of a specific working artist. What are the differences and similarities (color, warmth, balance, etc.) between the two types of images. Do the generative-image AI tools exhibit bias (culturally and aesthetically)? How?
Courses requiring students to to read and understand difficult texts
PDFs and texts can be input into a text-based tool and asked to summarize the text. Trying different prompts, students can assess the quality of the generated responses while deepening their understanding of the text assigned.
Course requiring students to create an annotated bibliography
Have students create an annotated bibliography on their own and prompt generative AI to make an annotated bibliography on the same topic. Ask them to compare the bibliographies and grade what generative AI has produced.
Courses requiring students to create original artwork
Some activities, courtesy of Noah Fisher, part-time assistant professor in Parsons:
- Have students create an image fragment, then ask them to extend it in an image generating tool such as DALL-E. Then edit the extended image.
- Use AI to create a moodboard.
- Generate backgrounds for an animation assignment.
Courses requiring students to interview people
Refine interview questions in advance by prompting an AI tool to play a specific role, and then interviewing the tool. Example prompt: Play the role of a flood warden with 20 years of professional experience and a personal experience with flooding. Analyze the output to refine interview questions before meeting and interviewing a real flood warden.
- Use text-based generative AI tools to create different types of assignments: Here are three examples for assignments for a design course generated by Bard and two examples for the same course generated by ChatGPT as well as three examples of assignments for a literature course generated by Bard and two examples for the same course generated by ChatGPT.
- Seven AI Assignments for Students, from Ethan Mollick
- 100+ Creative Ideas to Use AI in Education
- Daniel Stanford’s Substack
- Learn with AI (from University of Maine)