Reduce AI Cheating with Assignment Design

Reduce AI cheating with assignments designed with TRUST.

Transparency
Real World Connections
Universal Design for Learning
Social Knowledge Construction
Trial and Erro
r

Transparency. In addition to students’ usual questioning about why they have to learn something, now students add, why do they have to write an essay or do other assignments ChatGPT could do in seconds. Making the purpose and requirements for the assignment as clear as possible helps students understand the value of completing an assignment (beyond getting a good grade), and most will be less motivated to cheat with AI.

Top of Torrey Trust’s User Experience Research Project document illustrates making an assignment’s purpose clear.
Trust,T. (2023, August 4)

Real World Connections.  Make the assignment as applicable to the real world as possible. When students see an assignment is relevant to their lives or helps others, they are more likely to do the work.

Torrey Trust suggests several real world assignments:

Students could participate in a civic engagement project, design an open educational resource, build a working prototype of an invention, partake in a service learning activity, create a social media campaign, teach or tutor younger students, or address one of the United Nations Sustainable Development Goals.

In an Ancient History class, students could design social media videos to excite interest in the topics they are studying (see “Teens Are Going Viral With Theatrical History Lessons on TikTok”). In an Italian Studies class, students could create an open access eBook that teaches younger students about the Italian language and culture (see “Empowering College Students to be OER Creators and Curators”).

(See Real Life Learning Challenges for real world topics at all grade levels and various disciplines.)

Universal Design for Learning (UDL). The framework has three main principles: Multiple Means of Engagement, Multiple Means of Action and Expression, and Multiple Means of Representation. Using UDL can strengthen student interest, engagement, and motivation for doing an assignment, rather cheating by having AI do it.  UDL: A Powerful Framework describes UDL and how it strengthens learning, for all learners.

Social Knowledge Construction is about giving students the opportunity to deepen their understanding of class content through interactions with others, which can increase the relevance and value of an assignment. Constructing knowledge through connecting with others, isn’t just through collaborative group work. Reading text written by others, watching videos or presentations designed by others, communicating with others, and observing others – all are learning by connecting to others.

Trust makes assignments tap into social knowledge construction by having students invite others to participate in the assignment. For instance, in the User Experience Research Project mentioned above, students have to find three-five peers to conduct usability testing of an educational digital tool, and they present this data in their final report.

Another way to provide social knowledge construction is to encourage students to get feedback on their assignment from, or to share what they learned from the assignment with, individuals outside the class.

Trial and Error. Give students the opportunity to learn through failure. In Torrey Trust’s classes, if students fail part or all of an assignment, she gives them feedback on how to improve their grade and then gives them additional time to revise and resubmit their work.

For a large class, one way to incorporate “trial and error” is by having low-stakes quizzes that can be taken multiple times to demonstrate mastery of learning, rather than one high-stakes midterm and one high-stakes final exam.

This post is adapted from Trust, T. August 4, 2023. Essential considerations for addressing the possibility of AI-driven cheating. Part 2. Faculty Focus. See the original article for further details.

Leave a Reply

Your email address will not be published. Required fields are marked *