Automated Differentiation of Homework Tasks in Distance Learning: Integrating Artificial Intelligence and Genetic Algorithm

Given that homework submissions are conducted electronically, plagiarism in distance education has become a prevalent issue. To provide tailored assignments for different learners while ensuring balanced difficulty levels, teachers need to invest significant effort, which may not guarantee the desired outcome. To address this challenge, we propose utilizing Genetic Algorithm (GA) to generate differentiated assignments. Specifically, our research involves four steps. First, we upgraded and annotated an existing question bank using a combination of Artificial Intelligence (AI) technology and manual review. The annotations included identifiers (IDs), scoring rates, response times, and key points of assessment. Second, we designed a composite fitness function and determined parameters such as crossover, mutation, and elitism retention. Third, we developed a GA program using PyCharm and executed computations for six distinct user requirements. Finally, we analyzed and discussed the computational results from two perspectives: the iteration process of the fitness function and the quality of the generated assignments. The results indicate that the GA-based system effectively converges during the iterative process of the fitness function. It reaches relatively high fitness values within a reasonable number of iterations, demonstrating its efficiency in optimizing assignment generation. The quality of the generated assignments meets the predefined criteria well, with accurate alignment in terms of score, time, and knowledge coverage. The system significantly reduces the time and effort required from teachers while providing diverse and suitable tasks for students, thus offering a practical and reliable solution to the problems in distance education’s homework assignment process.