The Assignment Makeover section of the Teach@CUNY AI Toolkit provides guidance for thinking through how your assignments may or may not interact with generative AI, followed by a discussion of scaffolding learning and emphasizing process as starting points for considering AI’s impact on assignment design. It then covers how to do an “assignment makeover” with the goal of discerning and responding to potential AI uses by students as part of an assignment. The section concludes with an invitation for instructors to visualize their assignments from the unique and varied standpoints of their students.

Back to Basics 

The mainstreaming and commercialization of generative AI tools has made it difficult to overstate the value of well-crafted assignment design. Instructors might consider a “back-to-basics mentality” when it comes to redesigning assignments, paying careful attention to its sequence of activities and the capacity for generative AI to replace or enhance steps along the way.

In his recent Inside Higher Ed post — If ChatGPT Can Do It, It’s Not Worth Doing — John Warner moves the needle on this point: “If a large language model (like ChatGPT or its brethren) can generate a product similar to or better than humans on the same writing task, that writing task is not worth doing.”

In turn, Derek Bruff, author of Intentional Tech: Principles to Guide the Use of Educational Technology in College Teaching, advises instructors to answer the six-question framework below:

• Why is this assignment fitting for the course?

• What are the specific learning objectives?

• How might students utilize AI tools in this assignment?

• Can AI undermine the assignment’s goals, and how can we counteract this?

• In what ways can AI enhance the assignment?

• How can the process be made more meaningful and supportive for students?

Scaffolding Learning 

To scaffold an assignment involves segmenting its broader requirements into smaller and more manageable building blocks. Scaffolding informal exercises and low-stakes activities into a larger assignments frames learning as an unfolding process of growth and discovery, rather than a product or problem in need of streamlining or automation.

The Teach@CUNY Handbook elaborates further on the value of scaffolding in assignment design:

In assignment design, scaffolding refers to a process where assignments begin with a series of low-stakes (low grade impact) exercises which build up to a final, larger assignment. Scaffolding allows you to break down the component parts of a skill or assignment and to offer students the opportunity to check in and receive feedback at each juncture. In this way, scaffolding is both a planning and learning tool and follows a similar process to backwards design to consider what a student needs to know in order to complete a task or assignment. For many students, especially those who are not familiar with what goes into larger academic assignments, it is very important to make potential structures and processes for completing larger projects more visible. Modeling that building academic work is an iterative process can help students tackle more complex projects later on in their careers.

Scaffolding learning can offer students a framework of accountability that is often lacking in projects or assignments that require only a single submission. Encouraging students to leave traces of their learning process can also provide usable insights for instructors who have yet to update the design of their assignment to accommodate potential uses of generative AI by students.

As acts of revision and recursion become more visible over time, scaffolding techniques can affirm for students that knowledge production does not follow a linear trajectory or a straightforward march from not-knowing to knowing. Instead, it can foster a critical and reflexive habit of mind that unfolds over iterative cycles of reevaluation, refinement, and rethinking. This mindset is integral in the fast-changing landscape of generative AI, where students need tools and strategies to visualize themselves and their thinking in deliberative, ever-evolving terms.

Taking Activity Inventories 

Instructors tasked with remaking an assignment to accommodate AI can conduct what is called an “activity inventory.” This exercise involves making a list of every cognitive task required to complete the assignment, Remember when you were asked to make a exhaustive list of steps for how to make a PB&J sandwich in elementary school? That’s an activity inventory.

In turn, it asks instructors to kick the tires of an assignment to better understand its core activities, identifying which tasks can be automated or replaced, and determining opportunities for revising the directions or format of the assignment. Doing so opens avenues for instructors to improve their assignment’s design, while simultaneously calibrating it to realities of AI use in their class.

Stress Testing Assignments 

In a workshop on AI and the teaching of writing, Tim Laquintano and Annette Vee suggest instructors attempt their own assignments by using ChatGPT to reengineer its required activities in the same sequence expected of their students. By inputting the assignment prompts into the AI tool, instructors can assess how the tool approaches and completes the assignment.

Stress testing helps identify potential weaknesses or loopholes that students might exploit using AI, and therefore enables educators to modify the assignment to ensure it requires more space for critical thinking, metacognition, or reflective practice, which generative AI tools cannot easily replicate. The goal of this process is not only to maintain academic integrity among students, but to also remake assignments that allow instructors to understand and assess students for their genuine understanding of the topic, and to invite authentic learning experiences that exceed the classroom, not to mention the chatbot interface.

Visualizing Student Perspectives 

Like most people, students come to use generative AI for a variety of reasons. Surely there are students for whom generative AI is nothing but a shortcut. But instructors should not forget those who use these technology for learning support, who benefit from its adaptability and reach, its concision and clarity of expression, or even its capacity to translate vague or complicated assignments into something they can understand. Others may use it because they aren’t confident in their writing, or are being asked to produce work in a language other than their native tongue.

It helps to keep learner perspectives in mind when remaking assignments that accommodate tools like ChatGPT, given the diverse life circumstances of CUNY students. Stepping into their shoes can inspire new heuristics for how to stress test and mold assignments to the needs of the moment. Visualizing CUNY student perspectives can also promote an ethics of care in the classroom, which supports students in efforts to reconcile their own proximity to these tools and the complicated, busy lives they often lead.

Imagine the student crunched for time and faced with the two options: submit their paper after the deadline and fail the course, or submit an ersatz paper using ChatGPT to pass the class and advance to their degree.

What would you do? When is it acceptable to prioritize the convenience of technology over the labor of hard work? These are questions worth asking, no matter your discipline or pedagogy.