Cao Thi Van Giang¹, Cao Cu Giac^2,

This study explores the integration of STEM education and artificial intelligence (AI) in chemistry teaching to foster a sustainable pedagogical ecosystem. Current secondary school practices reveal three critical operational gaps: a deficiency in interdisciplinary integration methodologies at the lower secondary level, intense high-stakes national examination pressures driving technological risk aversion at the upper secondary level, and pervasive administrative formalism in teacher professional development. To address these challenges, the research combines conceptual framework construction with an exemplary case study of a chemistry STEM project entitled "Chemical Fertilizers and Smart Agriculture". This project is structured across three progressive technical layers: direct soil pH and nutrient analysis using IoT sensors in field environments (Layer 1); the integration of large language models (such as ChatGPT) as personalized learning co-pilots alongside virtual laboratories to optimize learning trajectories (Layer 2); and the process-oriented evaluation of 21st-century competencies via a Digital Portfolio platform (Layer 3). Based on these empirical findings, the study proposes a bipolar set of breakthrough solutions: activating teachers' internal "self-efficacy" through digital self-directed learning, and reforming the "transformative mission" of teacher education institutions through the development of smart campus laboratories and professional learning networks. Finally, the study recommends institutionalizing Digital Portfolios parallel to traditional academic transcripts to alleviate high-stakes examination pressures and establish a genuine symbiotic connection between teacher education institutions and secondary schools in the AI era.

STEM education, artificial intelligence (AI), chemistry education, teacher self-efficacy, digital portfolio