Research trends and students' perceptions about the impacts of Generative Artificial Intelligence in Higher Education: a mixed exploratory case study

The rapid evolution of Artificial Intelligence (AI) will play a crucial role in driving innovation and advancements within higher education in the coming years. However, this progress will also bring about a range of new challenges.
Given the immense potential of AI, especially the latest tools, in higher education, our research aimed to achieve the following objectives: 1) identify the current research trends regarding the impacts of generative AI tools in higher education, and 2) characterize the usage patterns and opinions of students in an academic setting towards Generative AI (Gen AI) and the available tools. To accomplish this, we intentionally selected two classes from the second year of Computer Science in two distinct higher education institutions and with very different profiles.
To explore the existing scientific literature, we conducted a bibliometric analysis using the VOSViewer software. We analyzed the data information of all documents indexed in the SCOPUS database that met our selection criteria.
We also conducted a mixed-profile questionnaire to understand students' generative AI usage and thoughts on its hazards and benefits in higher education. A thematic content analysis of the responses given in the open questions followed. Before analyzing the corpus, we used T-Lab to find the most common keywords. This was followed by Word Association and Sequence Analysis, which helped us uncover the relationships between the narratives most regularly used words and supported our thematic content analysis.
The mixed approach achieved through the combined use of bibliometric analysis and content analysis helped identify scientific interpretations on the potentialities and threats to higher education posed by Gen AI and compare them with those shared by students chosen for the exploratory case study. With the prominent categories that resulted from common perspectives, the study revealed domains of higher and lesser appropriateness of the published theory.