Beyond AI Assistance: Developing a Human-AI Collaborative Learning Model to Enhance Data Science Competencies

Abstract

The rapid adoption of generative artificial intelligence (AI) has significantly influenced higher education, particularly in Data Science programs where students increasingly utilize AI tools for programming, data analysis, and problem solving. However, unstructured AI utilization may lead to superficial learning and excessive reliance on automated outputs, potentially reducing the development of higher-order thinking skills. This study aims to develop and evaluate the Human-AI Collaborative Learning (HACL) model as a pedagogical framework that positions generative AI as a cognitive partner in learning processes. A mixed-methods sequential explanatory design was applied involving 120 undergraduate Data Science students, with 60 students participating in the experimental group through the HACL model and 60 students in the control group using conventional AI-assisted learning. Quantitative data were collected through questionnaires and project-based competency assessments, while qualitative data were obtained from observations, interviews, reflective journals, and learning artifacts. The results indicate that students implementing the HACL model achieved greater improvements in AI Literacy, Computational Thinking, Critical Thinking, and Data Science Competency compared with the control group. Structural equation modeling confirmed that HACL positively influenced learning outcomes through AI Literacy and higher-order thinking skills as mediating factors. Qualitative findings showed that structured human-AI collaboration promoted iterative reasoning, critical evaluation of AI outputs, reflective learning, and responsible AI practices. This study introduces the HACL model as a practical framework consisting of six sequential phases to support meaningful, ethical, and human-centered AI integration in Data Science education.