Fei-Fei Li: Human-Centered AI

Human-centered AI requires a multidisciplinary approach that integrates cognitive science, ethics, and social sciences to ensure technology serves human well-being rather than merely replacing labor. Vision serves as a foundational element of intelligence, as evidenced by the evolutionary significance of the Cambrian explosion and its role in early childhood cognitive development. By moving beyond large-scale supervised learning toward curiosity-based and few-shot learning models, researchers can address complex, real-world challenges like senior safety and healthcare delivery. Initiatives such as ImageNet provided the critical data infrastructure for deep learning, while programs like AI4ALL work to democratize access to these tools. Ultimately, the goal is to create invisible, assistive technologies that enhance human capabilities and address systemic issues, such as ambulance dispatch optimization and water quality monitoring, particularly within underserved rural communities.