Abstract
Science education stands at a critical juncture in the twenty-first century. As societies grapple with pressing challenges—climate change, public health crises, artificial intelligence transformation, and energy security—the capacity to produce scientifically literate citizens and innovative researchers has never been more urgent. Yet science education remains largely anchored to twentieth-century paradigms: content-heavy curricula, high-stakes assessments that reward recall over understanding, and pedagogies that position students as passive recipients rather than active inquirers. This conceptual paper argues that fundamental rethinking is required to align science education with the demands of innovation economies and democratic participation in technological societies. It critically examines three interrelated challenges: the persistence of what Michael Von Maltitz terms "broken proxies" in assessment, the equity gaps that pervade access to authentic scientific practice, and the disruptive potential—and peril—of artificial intelligence in science learning. The paper then explores emerging opportunities: the mainstreaming of active learning pedagogies, the integration of data science and computational thinking, the promise of phenomena-based and project-based instruction, and the potential for AI-augmented personalized learning when guided by robust pedagogical frameworks. Finally, it proposes future directions organized around four pillars: reimagining assessment systems, investing in teacher capacity and professional formation, embedding authentic research and innovation in curricula, and constructing ethical governance frameworks for AI in education. The paper concludes that the transformation of science education requires not piecemeal innovation but systemic reconceptualization—one that places inquiry, equity, and human flourishing at its centre.