Abstract:

In modern society, the integration of Large Language Models (LLMs), such as ChatGPT, has emerged as a turning point, especially within educational context. These technologies offer diverse benefits and redefine traditional learning paradigms. The objective of this investigation is to explore how students employ critical thinking skills across different stages of their interactions with LLMs. The research instrument employed for this study has been a survey conducted among students in the fields of Education and Computer Science at the University of Andorra. Through both quantitative and qualitative analyses, the research examines not only the frequency and perceptions of LLMs usage but also the thought processes, decision-making strategies, and problem-solving approaches that students employ when ensuring the reliability of the LLMs responses. While the data has been analyzed with consideration to the gender of the survey participants, it is important to interpret it through the lens of field-specific data rather than solely relying on gender-based data. Preliminary findings indicate a multifaceted result, with students demonstrating diverse approaches and perspectives. While some students employ critical thinking to evaluate the reliability and relevance of information generated by LLMs, others rely on these tools blindly. Furthermore, the study shows instances where critical thinking serves against potential biases or inaccuracies inherent in LLMs’ responses. This research contributes to a deeper understanding of how students activate critical thinking skills when using LLMs. Moreover, it underscores the importance of improving critical thinking competencies within educational curricula to empower students with a comprehensive understanding of the capabilities and limitations of LLMs, as well as how to effectively use them.

Keywords:

Critical thinking, Large Language Models, Higher education, Students confidence evaluation.