This study explores the intersection of human intelligence and Artificial Intelligence (AI) to design knowledge-in-use science assessments for supporting students’ deep science learning. In the context of evolving educational paradigms, it seeks to harness AI tools (GPT), to enhance knowledge-in-use assessment design, ensuring equitable opportunities for diverse learners. Anchored in the Next Generation Science Assessment and an evidence-centered design, this study aspires to harmonize AI's computational strengths with human expertise in assessment design. Drawing from an array of theoretical frameworks—Hybrid Intelligence System, Distributed Cognition, and Self-Regulated Learning Theory—the study underscores the multi-faceted and dynamic nature of knowledge-in-use and the symbiotic integration of human and AI.Employing a Design-Based Research approach, the study proceeds in three stages: (1) Iteratively training GPT models for effective designing knowledge-in-use assessments; (2) Gathering multidisciplinary expert feedback on AI-co-designed assessments; and (3) developing a domain-specific GPT-model for tailored assessment design that capture knowledge-in-use and address diverse student needs. Diverse data analysis techniques, encompassing thematic analysis, and descriptive statistics, such as heat map and scatter plot, are leveraged. Anticipated results spotlight an exploratory GPT model adept at creating tailored assessments resonating with diverse learning needs while emphasizing equity, adaptability, and inclusivity. This study holds the potential to significantly enhance the educational landscape by advocating a balanced approach where AI complements human expertise, paving the way for a progressive and inclusive future in education.
Read
