Generative AI in Education: A Framework for Leveraging Digital Tools in Latin American Classrooms

Generative AI (GenAI) is entering classrooms with both promise and uncertainty. This paper develops a framework that organizes instructional uses of GenAI into four functions: content preparation, explanation, practice and feedback, and motivation and exploration. Using evidence from Uruguay's Ceibal program, including platform data and observation from classroom sessions in a computational- thinking program, we show that patterns of use are not neutral. Female teachers and students engage more with tools for preparation and structured practice, while their male counterparts dominate exploratory platforms. These trajectories risk reinforcing inequality by channeling GenAI toward automation for some and augmentation for others. To test an alternative path, we design a randomized controlled trial that integrates a chatbot for explanation and feedback, explicitly structured for teacher mediation. The paper concludes with implications for Latin American education systems, and highlights that deliberate design and phased implementation are essential to ensure GenAI heightens inclusion rather than disparity.