Continuing from the previous Nature Select team's empathy assessment method, I also reviewed their MAPO training approach. The purpose of this training method is to enable the model to maintain a stable "empathy persona" during multi-turn interactions and provide relatively consistent responses.

In other words, the focus of this methodology is on the Agent's behavioral performance itself, rather than understanding the true psychological model, guiding human-machine interaction toward explicit "health/goodness."

Moreover, both papers lack an operational definition of "empathy." Based on the text, it can be understood as "the ability to demonstrate understanding, support, reassurance, and non-aggressive responses in conversations." This definition is clear but arguably not sufficiently deep, and the conceptual ambiguity does not facilitate researchers in truly addressing the structural issues of Agent human-machine relationships.

Original text: