The Vulnerability of Traditional Examination Forms in Distance Learning in the Era of Large Language Models (LLMs)

This study examines the impact of generative artificial intelligence (GAI) — specifically text-generating Large Language Models (LLMs) — on assessment methods in a digital learning environment, illustrated by a specific case from the field of finance. The widespread adoption of GAI in the educational environment calls into question the validity of established forms of remote assessment. The paper analyzes the vulnerability of a multi-component written assignment in a finance course through a benchmark experiment: reference answers generated by three text-generating systems are compared with real student works in a distance learning format. The results reveal moderate to high levels of similarity with LLM-generated reference answers among a significant proportion of participants. This trend is characterized by distinct stylistic markers, structural similarities across assignments, and indications of possible AI reliance. The findings are discussed in terms of their implications for the reliability of remote assessment and teaching practice, and practical recommendations for adapting assessment to the changing technological environment are proposed.