A Human-Centred Framework for Teacher Professional Development in AI-Supported Science Teaching

Artificial intelligence (AI) has moved from a peripheral innovation to an everyday presence in science education, where generative models, adaptive tutors, virtual laboratories, and data-rich simulations increasingly shape how learners encounter scientific knowledge. However, science teachers remain unevenly prepared to make pedagogically, ethically, and scientifically defensible decisions about these tools. Existing professional development often focuses on technical familiarity with AI applications, while giving less attention to the disciplinary practices of science teaching, such as inquiry, modelling, experimentation, data interpretation, evidence-based argumentation, and assessment. This article proposes a human-centred framework for teacher professional development in AI-supported science teaching. The framework is organised around five interlocking dimensions: foundational AI and data literacy contextualised in science; pedagogical content knowledge for AI-mediated inquiry; ethics, equity, and human agency; design competence for adaptive use and AI-supported assessment; and sustained, collaborative professional learning. Developed as a conceptual synthesis, the framework draws on the UNESCO AI Competency Framework for Teachers, recent extensions of TPACK into intelligent-TPACK and AI-TPACK, Human-Centric Artificial Intelligence Pedagogy, and emerging empirical studies on AI-supported classroom practice. It contributes by translating broad AI competency models into a science education-oriented professional development structure that foregrounds inquiry, modelling, experimentation, evidence-based reasoning, ethical judgement, and teacher agency. The article contributes to the literature by translating broad AI competency models into a science education-oriented professional development structure. It illustrates the framework through science-specific classroom scenarios and proposes a phased implementation roadmap from acquisition to deepening and creation. The article argues that effective AI-supported science teaching depends less on access to tools alone than on cultivating teachers who can critically evaluate, adapt, and humanize AI in the service of authentic scientific inquiry.