From Interaction to Relationship: The Role of Memory, Learning, and Emotional Intelligence in AI-Embodied Human Engagement

Abstract

The evolution of artificial intelligence (AI) has profoundly reshaped human-AI interactions, transitioning from rule-based systems to advanced machine learning algorithms capable of nuanced tasks. This paper introduces a novel architecture for embodied AI in human-robot interaction (HRI), designed to foster meaningful, long-term relationships. Central to the approach is the integration of persistent memory, enabling AI systems to recall past interactions, build continuity, and deliver personalised, contextually aware responses. Additionally, the architecture incorporates emotional intelligence to allow AI to recognise, interpret, and respond to human emotions, enhancing emotional engagement and trust. Key contributions of this paper include a unified approach that links attention mechanisms, memory, and generative AI to support dynamic, context-sensitive interactions; a focus on embodiment as a critical factor in HRI, highlighting its role in grounding interactions within physical and social contexts; and theoretical and practical advancements that extend existing attention-based systems by incorporating persistent memory and emotional intelligence for deeper human-AI connections. This work lays the foundation for developing AI systems capable of forming enduring, meaningful relationships with users, setting a new benchmark for human-centered AI design.