Undergraduate Students’ Perceptions of the Nature of Generative Artificial Intelligence (GenAI)-Influenced Science: A Structural Equation Modelling Approach

Generative artificial intelligence (GenAI) can facilitate scientific modelling, while it might produce biased scientific evidence. Hence, GenAI is influencing the ways of how scientific knowledge is produced; thus, students need to develop an informed understanding of how GenAI influences the characteristics of scientific works. To study this, we explore 182 undergraduate science and engineering students’ perceptions of the nature of science under the influence of GenAI. We design a questionnaire based on the family resemblance approach framework, which comprises aims and values, methods, practices, nature of knowledge, and social-institutional aspects. From descriptive statistics, we reveal that students have the weakest perception of the aims and values of using GenAI in scientific activities. In contrast, they have the most robust perception that GenAI can facilitate scientific practices. Structural equation modeling also reveals that students’ perceptions of the aims and values of using GenAI in scientific activities directly affect their perceptions of the nature of knowledge and practices influenced by GenAI. However, these dimensions do not significantly impact students’ perception of how social-institutional factors affect scientific knowledge in GenAI. We argue that teaching and learning in higher education should foster a linkage between students’ perceptions of social-institutional aspects and those of other aspects of the nature of GenAI-influenced science.