Empathy is at the heart of human relationships, enabling us to understand and share the emotions of others. But how can we translate such a deeply human trait into artificial intelligence? In this talk, I’ll begin by highlighting our contributions to this field, including user experiments that study emotional interactions, annotation studies refining emotional insights, and tools such as the MEED framework, PEACE model, and EDOS dataset, which have advanced empathetic response generation.
Building on these foundations, I’ll explore the psychological and philosophical underpinnings of empathy, focusing on how theories such as simulation theory and theory of mind can inform the creation of AI systems capable of true emotional understanding. We’ll examine key constructs of empathy—affective, cognitive, and compassionate—and discuss how these elements can move AI beyond surface-level interactions. We describe a future where empathy in AI redefines human-technology relationships, bridging the gap between artificial and general intelligence and fostering collaboration, trust, and a more compassionate digital future.