AI AND ACADEMIC INTEGRITY

AI presents tremendous opportunities for education, and as a College we are continually exploring how it can be used in the classroom to enhance, extend and engage in learning. However, the technology also presents significant challenges and one of these is the impact of generative AI on academic integrity (often termed AIgiarism), where work created by AI is presented as a student’s own.

In order to minimise the impact of AIgiarism, the College is implementing a number of measures, including clear communication of any permitted AI use as standard practice for all assessment tasks. We are also redesigning assessments to include more in-class writing, process portfolios and oral presentations. However, it is not practical for all assessment tasks to be AI-proofed, particularly those that are TASC-mandated work requirements, and for this reason we must have measures in place to respond to suspicions of AIgiarism.

We know that AI detectors have limited reliability and, unlike traditional plagiarism detectors which can link to a source, their results can’t be verified. Teachers may use AI detectors as a signal that AI has been used, but will not base an accusation of AIgiarism on this alone. Rather, suspicions that AI has been used inappropriately are usually based on disparity between work produced in class and work produced at home, or on a misalignment between the work submitted and the expectations of the task.

If a teacher believes that generative AI has been used inappropriately on an assessment task, they may put the task through multiple AI detectors in order to see consistent patterns and findings; ask current or past teachers to also review the student’s work; and speak with the student to ask them to explain their ideas and choices.

Despite the inherent limitations of AI detection, it is essential that teachers are able to trust the authenticity of student work. Because educators can’t prove that AI has been used, we are moving to a practice where students must take responsibility for proving that they have not used AI. The best way they can do this is by retaining all drafts, notes and AI prompts used in producing their assignments. All assignments must be completed while connected to the College’s OneDrive, using a Microsoft product that allows the version history to be saved. If students do not comply with these requirements, and are unable to prove that the work is their own, we must conclude that the authenticity of their work is in question and apply our Academic Integrity Policy.

We recognise that this is an unprecedented approach to suspected academic misconduct, but the rise of generative AI presents unprecedented challenges to the integrity of our results and grades. Our approach to academic integrity in the age of AI will be clearly communicated to students when tasks are assigned to enable their compliance with these expectations.

Kate Gard
Director of Teaching & Learning- Penquite