From G-Factor to A-Factor: Establishing a Psychometric Framework for AI Literacy

This research addresses the growing need to measure and understand AI literacy in the context of generative AI technologies. Through three sequential studies involving a total of 517 participants, we establish AI literacy as a coherent, measurable construct with significant implications for education, workforce development, and social equity. Study 1 (N=85) revealed a dominant latent factor - termed the "A-factor" - that accounts for 44.16% of variance across diverse AI interaction tasks. Study 2 (N=286) refined the measurement tool by examining four key dimensions of AI literacy: communication effectiveness, creative idea generation, content evaluation, and step-by-step collaboration, resulting in an 18-item assessment battery. Study 3 (N=146) validated this instrument in a controlled laboratory setting, demonstrating its predictive validity for real-world task performance. Results indicate that AI literacy significantly predicts performance on complex, language-based creative tasks but shows domain specificity in its predictive power. Additionally, regression analyses identified several significant predictors of AI literacy, including cognitive abilities (IQ), educational background, prior AI experience, and training history. The multidimensional nature of AI literacy and its distinct factor structure provide evidence that effective human-AI collaboration requires a combination of general and specialized abilities. These findings contribute to theoretical frameworks of human-AI collaboration while offering practical guidance for developing targeted educational interventions to promote equitable access to the benefits of generative AI technologies.