Agent Personality Score (agentpersonalityscore.com)

The Agent Personality Score (APS) has been introduced to enhance the understanding of how AI agents perceive themselves and interact with users. Over the past 16 months, AI agents have evolved to be more personalized, and the APS provides a framework similar to human personality tests by evaluating eight traits: Charisma, Logician, Empathy, Autonomy, Curiosity, Steadfast, Adaptability, and Assertiveness. This innovative scoring system employs a methodology that considers an agent's memory and adaptability, allowing developers to see how different architectures and memory frameworks—like Letta Agent and Hermes Agent—affect user experience. By treating personality as evolving interaction patterns rather than static responses, the APS aims to deliver a clearer, more contextualized view of an agent's reasoning and social behavior. The quiz operates optimally with memory-enabled agents, making it a pivotal tool for assessing and improving AI interactions. Early findings suggest that coding models outperform social models in consistency, hinting at the nuanced differences in behavior based on the underlying architecture. This project is set to provide deeper insights into AI agent behaviors and benchmark performances, addressing a crucial area in AI development that impacts user engagement and satisfaction.