This project outlines a new data-driven psychometric paradigm for measuring non-cognitive constructs, such as attitudes, traits, and values, in psychology and related fields. Addressing the current crisis of replication and construct proliferation, the project proposes leveraging advancements in digital data collection, artificial intelligence (AI), and natural language processing (NLP) to revamp traditional scale construction. By integrating AI into the stages of construct development—conceptualization, operationalization, and validation—the approach enhances precision and validity. One key innovation is random item sampling from larger item pools, allowing for more comprehensive construct representation and flexible measurement. The proposal aims to improve the reliability and validity of self-report measures, offering an efficient, scalable solution to longstanding issues in psychological measurement.