Large Language Models in Academia: a case study at the Technical University of Moldova

Abstract

Large Language Models (LLMs) have significantly impacted academic settings, presenting an innovative opportunity for learning and research, alongside challenges related to accuracy and ethics issues. To assess the effectiveness and acceptance of LLMs in educational environments, we conducted a comprehensive survey on 510 participants, including 470 students and 40 teaching staff, followed by a detailed exploratory data analysis of the results. The results show that LLMs exhibit parity in usage between men and women, while there is a strong imbalance in the frequency of LLM usage between students and teaching staff. Additionally, we observe a notable discrepancy in the daily usage rates, with 50% of students, employing LLMs on a daily basis, in contrast to only 18% of teaching staff. Both students and teaching staff predominantly use LLMs for searching information (21%), information analysis (20%) and code generation and editing (18%). Furthermore, the majority of participants express satisfaction with LLMs, with 91% of students and 85% of teaching staff recommending their usage. This high level of endorsement underscores the perceived value and utility of LLMs in academic contexts.