Abstract
Background:
The application of large language models (LLMs), such as Chat-GPT, DeepSeek, and Gemini, has been widely explored in recent years, including within healthcare education. While LLMs provide numerous potential benefits for students, concerns have also been raised that their use may impact students’ learning efficacy and academic performance. This study aims to investigate the attitude and usage of LLMs among Hong Kong dental students in their undergraduate studies, as well as factors affecting their acceptance of LLMs in dental education.
Materials and Methods:
This study employed a cross-sectional questionnaire survey, adapted from a previously validated Technology Acceptance Model (TAM)-based survey instrument (TAME-ChatGPT). Both electronic and paper-based questionnaires were distributed to Year 2 to Year 6 Bachelor of Dental Surgery students at the University of Hong Kong between November 2025 and February 2026. Data was collected on the frequency, purpose of use, and type of LLM used, as well as attitude and usage scores assessed from TAME-LLM. Key themes were identified in students’ short-answer responses. Bivariate analysis was performed to determine the associations between students’ demographic data, frequency of use of LLMs, and TAME-LLM scores.
Results:
The survey received responses from 145 students, all of whom had used LLMs to support their dental undergraduate learning. The results indicated a neutral attitude towards LLMs, influenced by high perceived risk and anxiety towards LLMs, a positive attitude toward technology, alongside a positive social influence. Usage of LLMs was positive, driven by high perceived ease of use, perceived usefulness, and positive behavioral factors. Bivariate analysis revealed that frequent use of LLMs was correlated with high perceived usefulness, positive attitude to technology, and positive social influence, but was not correlated to perceived risk or anxiety. A weak positive correlation was found between LLM usage and perceived academic performance. Students used LLMs most frequently in problem-based learning (PBL) contexts, and less frequently in didactic and clinical learning settings.
Conclusion:
LLMs were frequently used by dental students in their undergraduate studies, particularly in PBL contexts. High perceived usefulness, positive attitude towards technology, and positive social influence were the main drivers of LLM use, whereas high perceived risks and anxiety did not seem to deter students from using LLMs. When integrating LLMs into the dental curriculum, dental educators should guide students on appropriate and effective use of LLMs. Future efforts may focus on overcoming the current limitations of LLMs in clinical learning settings.