LLM for Question Generation and Validation to Enhance School Level Student Assessment

Recent advancements in large language models (LLMs) have accelerated the growth of several AI-based solutions in the daily lives of people, including in the field of education. Using LLM in education can potentially automate various aspects of the teaching-learning process. This work explores how LLMs can assist educators in generating questions for student assessment. Using sophisticated prompt engineering, a system has been developed to generate multiple levels of questions from a single topic, enabling teachers to evaluate and curate a variety of questions attuned to the diverse needs of learners. This innovative approach diverges from conventional ICT-based techniques that primarily rely on keyword matching and superficial metrics. The system generates questions, taking into account the semantic meaning and contextual relevance. Furthermore, LLM evaluation metrics have been applied to assess the methods used for the generation of questions, and the results are presented accordingly.