ChatGPT assistants in online higher education and student satisfaction: a case study

Abstract

The perception and satisfaction of students about a virtual assistant based on OpenAI ChatGPT 3.5, integrated in 21 different subjects of the virtual campus of an online university, have been analyzed in this study. Using a mixed methodological approach, information was collected on a sample of 391 students using the validated COMUNICA questionnaire, which included four constructs: Virtual Assistant Efficiency, Learning Impact, Skill Development, and Technical and Accessibility Aspects. The analysis included descriptive statistics, inferential statistical tests, Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA), complemented by a qualitative analysis of student and teacher comments. The quantitative results highlight that the female gender values the effectiveness of the assistant more than the male gender. The CFA confirmed that the factors can be grouped under a single latent variable: student satisfaction. In addition, the efficiency of the virtual assistant was found to be the most significant factor in the perception of student satisfaction, followed by the impact on learning, skill development and technical aspects. The qualitative analysis revealed mostly positive perceptions, highlighting the usefulness of the assistant in learning, an interest in extending its use to other subjects and suggestions for improvement in the accuracy of answers and functionality. It is concluded that virtual assistants have a positive impact on higher education, optimizing autonomous learning and educational interaction, although technical and design challenges persist that limit their full potential.