The ECOLE-MIRACLE project is ambitiously positioned at the forefront of conceptual learning, aiming to revolutionize how machines recognize and interpret images and videos in an open-world context. Instead of relying solely on predefined datasets, ECOLE harnesses the immense power of unlabeled videos and documents available globally, using them as a foundation for learning. Its distinctive methodology involves interactive curricula that evolve in complexity, allowing the system to gain nuanced insights into the attributes and affordances of objects. Furthermore, it comprehensively understands activities by discerning the conditions under which they occur and recognizing their participants. ECOLE seeks to establish a comprehensive, global-scale understanding of objects and events, bridging the gap between raw data and meaningful recognition.