HAI-MinePy: A Human–AI Co-Creation Model for Developing Computational Thinking and Python Coding through Minecraft Education

Abstract

Generative artificial intelligence is reshaping coding education by enabling learners to obtain code explanations, translations, algorithms, and debugging support through conversational interaction. However, the pedagogical challenge is to ensure that AI support strengthens, rather than replaces, students’ computational thinking and creative agency. This study developed and examined the HAI-MinePy Model, a Human–AI co-creation model for Minecraft Education Python coding. The model positions ChatGPT as a co-creative scaffold that supports MakeCode-to-Python translation, algorithmic brainstorming, concept explanation, and debugging hints, while students remain responsible for problem framing, testing, revision, and final artefact explanation. Grounded in constructionism, computational thinking, scaffolding theory, AI literacy, and human-centred AI, the model was implemented through seven iterative phases: problem framing, prompt formulation, code translation, algorithm co-design, human debugging, code refinement, and reflective evaluation. A mixed-methods design-based research approach was employed with 32 students across eight instructional sessions. Data were collected through computational thinking tests, Python coding tasks, prompt records, debugging logs, Minecraft artefacts, observations, reflection journals, and interviews. Findings showed substantial improvements in students’ computational thinking and Python coding performance, particularly in loops, decomposition, algorithmic thinking, debugging, and code explanation. Prompt records indicated that students increasingly used ChatGPT for translation, explanation, and debugging hints rather than direct answer generation. Qualitative findings further revealed that Minecraft functioned as a visible testing environment, while debugging became the central mechanism through which Human–AI co-creation became educationally meaningful. Nevertheless, early tendencies towards AI over-reliance highlighted the need for teacher scaffolding, prompt rules, and process documentation. The study concludes that the HAI-MinePy Model offers a theoretically grounded and pedagogically practical framework for integrating generative AI into coding education while preserving student agency, creativity, accountability, and responsible AI literacy. It also contributes to emerging scholarship on Human–AI co-creation by demonstrating how learner-centred AI use can be systematically designed, assessed, and refined.