Mechanisms of enhancing learning with unequal preparation: An experimental study on generative artificial intelligence use and proficiency in programming learning

Prior learning preparation, encompassing learners' prior knowledge, resources, and readiness for learning new material, plays a critical role in a new learning process. Unequal learning preparation commonly exists among learners with varying socioeconomic and academic backgrounds, and can further exacerbate the educational divide. By reducing learning cost and providing less prepared learners with personalized delivery of information and knowledge, Generative Artificial Intelligence (GenAI) may help mitigate disparities in learning preparation. In this study, we investigate the roles of learners' use of GenAI in their learning processes, considering varying levels of learners’ prior learning preparation and proficiency in GenAI use. Specifically, we examine the effects of the use of GenAI tools on learning outcomes through its impacts on perceived informational benefit, learning cost, and knowledge fit in the focal learning process, moderated by learning preparation and GenAI proficiency of learners. Based on an experiment in a programming learning context, we find that although GenAI use significantly reduces learning cost, especially for less prepared learners, the cost reduction effect has not been translated into improved learning outcomes. We find no significant moderating effects of learning preparation on how GenAI use affects informational benefit and knowledge fit, further showing that GenAI tools have not yet been effectively utilized to help less prepared learners or reduce the educational divide. As indicated by the significant moderating roles of GenAI proficiency, to fully leverage the power of GenAI, improving GenAI proficiency can be crucial to ensure learning effectiveness for learners with different levels of learning preparation.