Generative AI presented a direct challenge to the integrity of my final-year brand management assessments. Students could produce plausible, structured brand audits using AI tools without engaging in the research, analysis or critical thinking the assessment was designed to develop. Reactive institutional responses such as updated plagiarism policies and detection software, address the symptom rather than the cause, and do little to preserve the pedagogical value of the assessment itself. The real challenge was not simply preventing misuse, but redesigning assessment so that genuine learning remained the most effective route to a good grade. The question: how do you make AI misuse counterproductive rather than merely prohibited?
I redesigned the brand audit assessment for a final year brand management module, around a two-stage structure that embedded Gen AI tool Claude directly into the learning process. In the first stage, students used both Claude Pro and free generative AI tools to produce comparative brand audits, actively interrogating what AI could and could not do. In the second stage, they produced original analysis drawing exclusively on paywalled academic databases and real-time industry sources (Mintel, Statista, peer-reviewed journals) that generative AI cannot access. Submitting AI output uncritically became self-defeating; the assessment required students to demonstrate precisely where AI fell short. Seminars were redesigned in parallel, incorporating live prompting exercises using Claude as an explicit, assessed competency. Students refined their prompting techniques iteratively across the module, developing a critical and purposeful relationship with AI as a professional research tool rather than a content generator.
The decision was driven primarily by professional responsibility. My students are final-year undergraduates entering a marketing industry where AI fluency is already an employer expectation. Designing assessments that simply excluded AI would leave them underprepared for the roles they were about to take on.
The approach was also shaped by a recognition that assessment design, rather than policy enforcement, was the more durable solution. If the assessment itself required students to go beyond what AI could produce, the integrity of the learning experience could be preserved without relying on detection tools of questionable reliability.
The evidence base is practitioner-led. Post-module survey data from 22 voluntary respondents (cohort of 150) indicated that 86% agreed the assessment taught them to use AI as a research tool rather than a content generator, and 91% felt more prepared to use AI responsibly in a professional context.
Outcomes were broadly positive but require honest qualification. Survey data indicated encouraging results across all four quantitative measures, with no students reporting reduced confidence and fewer academic integrity violations were recorded across the full cohort. However, the mark scheme allocated too much weight to the generative AI component, which inadvertently disadvantaged some students whose foundational brand audit skills were underdeveloped as a result. This was an unintended consequence of the design. The mark scheme will be rebalanced in the next iteration to ensure AI remains a research tool within the assessment, not an assessed endpoint.
The most valuable insight is that AI integration must protect the cognitive journey, not just the final output. Students can produce sophisticated work without genuinely understanding it, and if assessment design allows that, the learning is lost, regardless of the grade awarded. The goal is not to exclude AI but to ensure its use requires genuine thinking. Critically, mark scheme weighting matters: allocating too many marks to AI-facing components can inadvertently undermine the core disciplinary skills the module exists to develop. Start with one assessment, be transparent with students about your rationale, and build in structured reflection to evaluate what is and is not working.
Do you have examples of how you have used AI in your education, research or leadership roles within higher education? Please do submit your short case-studies to us
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.