In-depth exploration of LLM’s adoption in shaping future expectations among Sri Lankan undergraduates

Abstract

An extensive language model called ChatGPT created by the OpenAI Foundation, has created a significant impact in academia and industries, especially due to its impressive performance, particularly in synthesizing human-like results from vast amounts of data. Despite numerous research efforts on top of ChatGPT, there is a paucity of literature providing in-depth analysis among undergraduates, the group that utilizes ChatGPT the most. Thus, this study addresses this gap through an in-depth analysis based on data collected through a structured questionnaire administered to 171 undergraduates. The questionnaire covered various aspects related to the usage of ChatGPT, including awareness, frequency of usage for learning, accuracy of information, perceived improvements, and future expectations from language models. Results indicate that ChatGPT is well-known and used frequently for academic reasons, especially for preparing assignments and exams. While most students appreciate LLMs, appropriately half have encountered partially accurate answers, revealing limitations in information accuracy. Some students emphasize the importance of promoting individual study and critical thinking skills, while others advocate for more comprehensive responses, including visuals and video lectures. Many students claim that LLMs have improved their academic performance, and LLMs tend to be favored among these undergraduates. These insights underscore the potential for LLMs like ChatGPT to transform learning and research experiences, emphasizing the need for ethical surveillance and adaptation to shifting student expectations while addressing issues of accuracy and bias. This positions LLMs as useful tools for knowledge acquisition and exploration in academia's changing digital landscape.