An artificial intelligence meta-skills framework for undergraduate research

Undergraduate research faces unprecedented challenges in the digital era, as exponentially growing data increasingly shapes the university ecosystem. However, artificial intelligence, particularly tools such as ChatGPT, offers significant opportunities to address this transformation. Effective use requires advanced competencies in generating, acquiring, storing, and analyzing data. In response, this study developed and validated a new Big Data Socio-Technical Competence Scale to assess undergraduate students’ competencies in research contexts. This study examined the relationship between Big Data socio-technical competencies and undergraduate students' academic use of ChatGPT, as well as the underlying meta-skills for using AI in research. A sequential explanatory mixed-methods design (QUAN → qual) was adopted, involving 418 undergraduate students. We used two measurement instruments: the newly developed and validated Big Data Socio-Technical Competence Scale and a scale measuring the academic use of ChatGPT; both showed high internal consistency, along with a semi-structured interview protocol. The findings show that Big Data competencies were generally low, whereas the academic use of ChatGPT for research was predominantly medium to high. Structural equation modeling revealed that socio-technical competencies in Big Data positively and significantly predict the academic use of ChatGPT (β = 0.269, p < .001). The qualitative phase identified nine Artificial Intelligence Meta-Skills for Undergraduate Research: (1) Critical Data Literacy, (2) Critical Thinking, (3) Epistemic Data Management, (4) Evidence-Based Decision-Making, (5) Critical Data Communication, (6) Metacognitive Regulation, (7) Emotional Self-Regulation, (8) Responsible Human–AI Agency, and (9) Ethical Reasoning. The study discusses educational implications and practical applications.