Designing AI Aware Assignments

Share with students why they are completing a certain assessment and how it connects to the course learning goals or helps students develop the skills they will need for their future learning or professional life.  

Student questions addressed by the purpose component include:  

• What skills or knowledge will I gain from completing this assessment?
• Which course learning goals does this assessment address?
• How will I be able to use this knowledge or these skills in the future (e.g., in courses or in my career)? 

Share with students the specific actions they should take when completing the assessment. This can include specific steps to be taken in a certain order, recommendations for approaching the assessment, and/or common mistakes for them to avoid. If the assessment is designed to elicit productive struggle among your students, you might explicitly explain this. For example, you might say, “This assessment is designed to be difficult; however, I know you are capable of completing it. While I expect you will likely struggle with the assignment as you do it, I am always happy to provide additional assistance if you attend office hours or send me an email.” 

Student questions addressed by the task component include:  

• What specific items need to be submitted?  
• How do I complete the assignment? How can I avoid common mistakes?
• When is this due?
• Where can I do this work? Do I need the internet or library? Where do I submit this work (Canvas)?
• Who do I work with for this assignment (e.g., do I need to work alone or with a group)? Who else can I seek support from to complete this assignment (e.g., librarian, community member, tutoring center)?  

Share with students the characteristics of the final product and how it will be assessed. Consider providing a rubric and/or multiple exemplars to show students the range of what they could produce and prevent them from relying too heavily on one approach.  

Student questions addressed by the criteria component include:  

What are the parts of the assessment? What part matters most? 

• How will I know what’s expected? 
• How do I know I am on the right track? How will I know I’m doing good work? 

To see specific examples of the TILT transparency framework applied to assignment prompts across disciplines, review resources on the TILT higher ed website.  

Discuss with students the importance of practice and highlight how the assignment is an opportunity for practice that will apply to future assessments or skills. For example, problem set questions, particularly when written in a similar format as exams, are great practice and preparation for those (graded) assessments. Students may not always recognize this, so it can be important for instructors to explicitly point it out. Information about the value of the assessment and why a particular assessment is being assigned could be discussed when initially sharing the assignment with students, or could fit well within the “purpose” section of the TILT transparent assignment design template. 

Help students take intellectual responsibility for their submitted work. Students who complete their own work should be able to defend the arguments they use in essays or the theorems they reference in problem sets. One way to do this is through oral examinations. In a course where take home essays or exams are assigned, students complete and submit their work. The instructor reviews and grades the work, and notes areas to probe student thought more deeply – Why did you choose this hypothesis or thesis statement? How does your work connect with the broader literature? How do you address this complication or objection? Then, during brief oral exams, students would be asked to respond to these questions. The goal is not to “catch” the students using AI, but instead, to further assess their approach to formulating their thoughts, arguments, or processes in an AI-free setting.  For more information on this approach, review Matthew Hammerton’s essay. 

Ask students to incorporate examples, theories, or discussions from class as part of their assignment. This technique can mitigate AI use by asking students to engage more deeply with aspects of the course that an AI tool does not have ready access to. Students could write a “discussion response” after class summarizing the main points of the discussion, analyzing their peers’ contributions, or suggesting areas for further class inquiry. This has a secondary benefit of encouraging students to be more present and attentive during class sessions.  

Similar to situating assignments in the course context, consider using location-based assessments, such as asking students to attend museums, labs, cultural events, or seminars and analyze the connections to the course content. You could also ask students to collect field- or location-based data. For example, in a sociology of gender course, students could take field notes on how children of different genders interact on a playground and then analyze that data. In a microbiology course, students could take samples from areas around campus and analyze the microbes present in a lab session. If you are interested in incorporating museum visits or collections into your course, consider reaching out to the Feitler Center at the Smart Museum of Art for guidance. For experiential learning within the city of Chicago, Chicago Studies can help you facilitate that experience. 

Ask students to document their learning process in a way that circumvents AI use. For example, ask students to explain their reasoning by adding short written reflections to your assignments, e.g., “In a few sentences, explain why you chose this approach to solve this problem and which key concepts from class it illustrates.” Alternatively, students can explain the learning process they used to develop a final product via student presentations, oral exams, video/audio recording, etc. For a full-term project, students could compile “learning logs” or reflective notebooks where they are asked to make a certain number of entries per week related to course content, questions, applications, assessments, or even the students’ own reflections. 

Process tracking is an approach that allows instructors to gather data about how a piece of writing was formed and is gaining traction among some educators to address unsanctioned AI usage. Programs such as Google Docs, Process Feedback, and Grammarly Authorship, can track how much time students spend on a task, identify text that is copied and pasted, and note sections of text that were typed by the student, among other features. These programs allow instructors to “see” more of their students’ writing process. Some may be concerned that this approach amounts to intrusive surveillance of students; however, if the information is used to help students reflect on their writing process it may be a helpful tool. Requiring documentation of the writing process may change how students approach writing and affect the level of trust in the student-instructor relationship. 

Many instructors are moving toward in-class assessments and requiring students to handwrite essays or exams, to ensure they are not using artificial intelligence. This is tempting; however, faculty should be cognizant of the fact that many of today’s students have little experience with handwriting for long periods of time. It’s important to consider the difficult task of deciphering student writing and the additional logistical load this may create, not to mention the accessibility of such practices. One way to address the accessibility concerns around handwritten assessments is through using Respondus LockDown browser. This browser is integrated into Canvas and prevents students from accessing other web pages or programs on their computer while it is in use. Learn more about the Respondus LockDown browser from from ATS here.  

A flipped classroom is one where direct instruction occurs outside of class, such as by asking students to watch lecture recordings or engage with readings prior to class. Instructors can then use class time for applications of the material, guided practice, problem solving, and discussions. In this way, students would essentially be completing their “homework” while in class, which can discourage AI usage since support from their instructor and peers is readily available. 

Students can compare AI content to their own experience of engaging with a reading, for example, or critique how an AI tool approached a research project compared to how they might have approached the same project. This can help them build information literacy skills and become more discerning users of AI technologies, especially when asked to identify tasks at which the tools are effective and ineffective. 

For example, students could roleplay a business “pitch” meeting, practice explaining a concept to an AI tool that is acting as a 5-year-old, or use the AI as an interlocutor to explain their own expertise.  

Consider the skills that students might need for the course that aren’t essential parts of your objectives and how an AI tool might supplement those skills. For example, in a course on evolution, students could use AI to generate pictures of “missing links” and compare those to the actual fossil record, removing the need for artistic skills. Assessing students’ drawing skills is likely not a learning goal for this course, so there is less risk of circumventing the essential learning of the course when students use AI for this more creative application. Benjamin Morgan (Associate Professor, English Language and Literature) discusses this approach in his CCTL Teaching Spotlight, as does Lisa Rosen (Associate Senior Instructional Professor, Committee on Education) in her CCTL Teaching Spotlight.  

Many instructors find that students have gaps in their relevant prior knowledge for courses. For example, many students in introductory calculus classes need additional support around various precalculus concepts. Custom chatbots, developed by instructors and prompted by course materials, can serve to fill this gap and help support students. Read more about creating supportive custom chatbots in this ATS blog featuring Selma Yildrim (Associate Instructional Professor, Department of Math). 

In this framework, developed by the Learning Lab at UC Davis, students complete peer review of writing assignments (without AI), prompt an AI tool to review the same drafts, reflect on both types of feedback, and then revise. This format allows students to critically examine the value of both peer and AI feedback and utilize both in more effective ways to improve their writing. This approach could be expanded outside of writing classrooms to other projects students might complete, such as research proposals. For more, including prompts used to provide the most effective feedback, see the UCDavis PAIRR website. 

When students use AI in any way, it is helpful to ask them to reflect on the experience of using AI and where the tools succeed and fall short. You could prompt students to consider: how did the AI tool support and/or detract from my learning? For which tasks was the AI tool effective or ineffective? What considerations (e.g., bias, hallucinations, energy usage, cognitive offloading) did I keep in mind when using AI for this assessment? 

Some students may object to the use of AI tools, whether for ethical, environmental, or other reasons. If you are asking students to interact with a chatbot to critique its output, you could provide students with the output, for example, without requiring students to use the chatbot. If students are using AI for tasks outside of the necessary course skills, consider whether students can complete the assessment without the AI component and still meet the learning goals. If students are interacting with the chatbot for conversational purposes, could they do the same action with a peer? It is helpful to return to your course goals as you consider alternative assessments to ensure all students are practicing the requisite course skills.