Attention To Emotion Capabilities Is All You Need: The Artificial General Emotional Intelligence (AGEI). Suggestion for A Framework for Understanding and Regulating Advanced Emotional Capabilities in Artificial Intelligence (AI) Systems.

The evolution of artificial intelligence (AI) has progressed from narrow systems intended for specialized tasks toward the ambitious objective of Artificial General Intelligence (AGI) - single systems demonstrating human-equivalent or superior performance in all cognitive tasks, across domains and a wide range of environments. AGI represents a critical conceptual milestone, shaping research priorities, regulatory approaches, and ethical considerations within the field. Currently, the evaluation of AGI primarily focuses on computational and logical abilities, using benchmarks designed to assess skills like reasoning and problem-solving. However, this approach systematically neglects important socioemotional dimensions.The capacity to comprehend and respond appropriately to intentional mental states - to mentalize - is a crucial component of human psychological development, social learning, and wellbeing. Omitting such socioemotional dimensions in AI evaluation creates a conceptual asymmetry, particularly as current state-of-the-art systems increasingly exhibit advanced emotion recognition capabilities.