ChatGPT is one of many new, AI-based tools that can be leveraged to do a number of tasks, and has made its way to the forefront of conversations in teaching and learning conversations across the country.
Detection tools are in their infancy and while instructors are right to be concerned about the potential for increased academic dishonesty, there are no clear and concise answers for dealing with AI-generated work. That said, we do have some recommendations:
Strategies for detecting AI-generated work
- Compare student work to prior work. This is most easily done by utilizing EWU’s current plagiarism detection software (TurnItIn) which compares student work to a database of sources, including previously submitted student work. Plagiarism detection companies are also working on specific methods of detecting AI-generated work; however, due to the speed at which these tools evolve it is questionable how effective these tools will be.
- Compare student work to a set of sample answers from AI-based tools. AI-based tools such as ChatGPT, Chatsonic, and Chinchilla are open and available to all of the public, not just students or paying subscribers. One strategy for detecting AI-based work would be to ask AI questions yourself and see what answers it provides. You can then take this knowledge of likely responses and use it when reading/grading student work.
- While grading student submissions, look for:
- A lack of original ideas.
- Circular or repetitive ideas/logic.
- Critical thinking/analysis.
- Correctly cited work, including checking works cited to ensure the veracity of sources.
- Understanding of context and nuance, including understanding cultural references, idiomatic expressions, and sarcasm.
- Self-reflection and self-awareness; understanding of personal biases or motivation.
AI-generated work can be relatively sophisticated depending on the task and the directions given by the user, so detecting it can be difficult. Instead of relying on detection, instructors may need to respond with other changes to their course and/or pedagogy.
Additionally, it may be helpful to consider where in the curriculum AI-based tools may be an opportunity for student learning: utilizing it as as study “partner”, getting help with simple problems/tasks, evaluating responses for critical thinking/circular logic/false information/etc, or generating possible areas for future inquiry may be potentially helpful uses of this new technology for you and your students.
What instructors can do to discourage students from using AI-based tools improperly
- Make sure that students understand what constitutes academic dishonesty – Provide clear guidelines and expectations for academic integrity, the correct use of AI and other learning tools, and clearly outline the consequences for academic dishonesty.
- Foster a sense of community and stress the importance of learning objectives/tasks – Creating a classroom culture that values original thinking and creativity over rote memorization can create a sense of personal investment that will encourage students to be more active and interested in their learning. Adult learners highly value understanding why they are doing/learning something and want to know that it will have a direct impact on their academic, personal, and professional lives; making clear connections between course content and its importance in course and weekly objectives, lectures, and throughout your Canvas course will make it easier for students to make these connections and invest in your course.
- Focus on the whole learning process – Encouraging students to work on a large assignment over a longer period of time can be helpful in identifying AI-generated work, can reinforce the importance of student’s learning, and can ease the time and grade pressure of important assignments. It also helps to build a sense of instructor presence and increases student engagement in online learning while giving instructors an opportunity to provide regular feedback to student work.
- Ask open-ended questions – Open-ended questions require students to demonstrate understanding and apply their personal and academic knowledge to a problem; tasks AI currently struggle with. Additionally, asking students to draw connections between the learning material and their own lived experiences can be another way to increase student engagement while providing a way to identify potentially AI-generated work.
- Utilize best practices of instruction – Some teaching practices will likely need to be adapted to the incorporation of AI-based tools, but utilizing already common best practices such as: leveraging program, course, and weekly learning outcomes to inform students of the importance of what they are learning; utilizing multiple assessment strategies; focusing on higher-level thinking skills; incorporating group discussions and assignments; and including peer review assignments can help mitigate the effect of AI-based tools on your course while also generating positive student engagement, boosting student learning, and positively impacting student retention.