Harnessing the Generative AI Wave Towards Fair and Diverse Higher Education Assessments: A Comprehensive Analysis through an Innovative Lens of Students

Abstract

While Generative Artificial Intelligence (GAI), particularly tools powered by Large Language Models (LLMs), offer benefits in teaching and learning, they also raise critical concerns about academic integrity, fairness in examinations due to their potential for generating educational content. This evolving landscape requires higher education institutions to rethink their assessment models, ensuring they remain robust, inclusive, and aligned with the realities of AI-enhanced learning environments. In this backdrop, this study investigates the practical, GAI-resistant assessment frameworks in higher education. It explores how alternative, skill-focused methods such as oral exams (vivas) and AI-integrated tasks can be included in future assessment models. Central to the study is the understanding of how students perceive current assessments and envision future methods that fairly and effectively measure both knowledge and skills. The empirical investigation is based on a case study at a Swedish University. Research methodologies include a survey questionnaire administered to 30 students enrolled in a semi-theoretical course on innovation and technology, and a future workshop (FW) with 22 of them in five groups. The two research instruments corresponded to answering the two research questions, respectively. The survey results revealed students’ clear concerns about the academic integrity challenges posed by essay and report-based take-home assessments, as well as online quizzes. They also expressed apprehension about the potential impact of relying solely on proctored and supervised exams, highlighting the risk of reducing diversity in assessment methods, and thereby raising red flags for the need for a new and innovative approach to assessment methods that is hardly affected by unauthorised assistance from GAI. Responses to open survey questions reflected their problem-solving mindset and deep thinking of how cheating can be minimised by increased peer collaboration and solving real problems, contextualised to specific and ongoing learning activities in class. The outcomes of the FW provided insights, such as active learning-based assessments, combined with real-world problem-solving or context-specific question-based assessments. These findings are intended to inform course design, policy-making, and broader discussions on educational reform in the digital age.