Heuristic for Designing Learning Experiences with Generative AI

In other words, what is the target course learning outcome for the learning experience? Identify the “verb” that you most want students to do during this learning experience (e.g., debate, dissect, dance, dream, direct) so you can design with that action in the forefront of your mind.  

For example, consider an assignment in a public policy class that asks students to create a Public Service Announcement (PSA) for TikTok to educate Evanston teenagers about mental health and local resources available to them. This learning experience creating a PSA directly maps to the following student learning outcome in the course: “By the end of the course, students will be able to translate scientific research into audience-specific public health messaging.” The key verb for this learning experience is “translate.” Students are creating a PSA in order to practice translating mental health research to a specific audience, Evanston teenagers.

Brainstorm any concepts, terminology, or skills that students should have already encountered or developed before this learning experience, whether as part of prior learning experience in your course or prior to taking your course. 

Continuing with the example of the public health course, making a TikTok video requires some knowledge of how the app work as well as how to record, edit, and add music to a video—elements the instructor may not have planned spending too much time on in class based on the assumption that students make videos all the time in their personal lives. But, some students may not start with the technical know-how needed to create a video and may end up spending more time than intended learning the technology, which then takes away from the time they could have invested in reading scientific articles on mental health, researching local mental health services and Evanston teen demographics, and then applying that research to writing a creative script for their target audience. In addition, the PSA assignment might also assume that students already know how to find, read, and comprehend dense scientific papers, find demographic data, and write within the genre conventions of a PSA. Perhaps some of these skills will be addressed in the design of the course, but that may not be enough to get all students to where they need to be to synthesize these skills into a PSA video that effectively communicates to the target audience.

Generative AI tools can supplement moments when students may not have access to certain skills that will benefit the learning experience but not central to it. Do they need to be an artist, graphic designer, composer, or writer to participate in the learning experience or can a generative AI tool be their collaborator? 

For our PSA example, how a message is delivered is often just as important as what the message is in terms of overall effectiveness. But, if the course is more about developing public health expertise than public speaking skills, some students’ PSAs may be at a disadvantage if public speaking is not already a strong suit, which can be particularly challenging for English Language Learners and some neurodiverse students. As an option for any student, a generative AI tool could generate a voice recording of the text. Maybe picking teenagers’ voices that match the local demographics would even help with the effectiveness of translating the mental health message to the target audience. Alternatively, students could hear how an AI reads the script and then take notes about how the student wants to perform the script—where do they want to put emphasis, add emotional color, slow down, speed up, etc.

How might students test their newly acquired expertise on a topic by analyzing something created with a generative AI tool? How might generative AI give students options in demonstrating their learning?

For example, after students have read and digested peer-reviewed scientific papers on mental health and teenagers in the public health course, they could ask a generative AI tool questions about mental health from the perspective of an Evanston teenager and then compare the AI tool’s answers to what they have learned through their independent research. Where does the AI tool get it right and wrong? What does it include and exclude? How might knowing what the AI tool leaves out or where it may misinform influence the focus of their PSA script?

Asking students to reflect on their learning—what decisions they made, what they tried, what they would apply to a similar future learning experience, and why—is one of the most effective pedagogical methods for helping students recognize what they have learned and then for transferring that learning to new contexts. Generative AI tools can provide a mirror to students, whether through summary or visual metaphor, of what it sees in their work. Students can then use that reflection to self-assess if they think it is an accurate representation or a distorted funhouse mirror.

Finally, for the example of the PSA assignment in the public health course, students could use a generative AI tool to create an image that either captures the essence of the PSA or that represents their learning journey in creating the PSA. Students could respond to a set of prompts in a “cognitive wrapper” activity explaining why their selected generated image is an accurate reflection of their content or experience and what steps they took to collaborate with the generative AI tool to create that image.