From *Guiguzi* to AI Agents: How the Interactive Cognition Framework Reshapes the Reasoning Paradigm of Artificial Intelligence

This article introduces the open-source project "interactive-cognition", which integrates Eastern wisdom from Guiguzi and The Art of War, shifting the focus of AI reasoning from internal analysis to modeling others' cognition and managing information flow, thus providing a new paradigm for multi-agent interaction. This framework addresses the shortcomings of traditional AI in scenarios like negotiation and collaboration, promoting agents from "response machines" to "interactive participants."

Traditional AI focuses on internal reasoning (parameter count, architecture, data) and struggles in multi-agent interaction scenarios (negotiation, game theory). The vertical and horizontal strategies from Guiguzi (such as opening/closing, reflection, inner connection) and the game thinking from The Art of War (know yourself and your enemy, all warfare is based on deception) provide practical wisdom for solving problems of "understanding others" and "strategic interaction."

1. Modeling Others' Mental States: Maintain state vectors of one's own and the opponent's beliefs, intentions, capabilities, etc., to achieve recursive reasoning ("I think he thinks I want...");
2. Information Flow Management: Strategically choose ways to disclose/hide information to maximize the benefits of interaction goals;
3. Dynamic Strategy Adjustment: Update strategies in real time based on the opponent's behavior to adapt to long-term interaction evolution.

• Multi-agent Negotiation: Achieve more win-win outcomes through modeling opponents' preferences;
• Human-Agent Collaboration: Understand users' unexpressed intentions and provide precise assistance;
• Strategy Games: Combine psychological tactics to surpass the advantages of traditional probability calculations.

The framework adopts a modular design, which can be superimposed on LLM/RL agents like GPT and Claude, providing standardized interfaces. The open-source project supports extensions: introducing multi-cultural strategies (such as Machiavellianism, Confucian ethics) or domain-specific modules (diplomacy, medical communication).

Contemporary AI is influenced by Western rationalism and emphasizes formal reasoning; Eastern wisdom provides a context-sensitive and strategically flexible practical paradigm. This framework supplements the tradition of "unity of knowledge and action", helping AI to be effectively applied in complex social situations and promoting the development of cultural intelligence.

Challenges: High computational complexity of mental state modeling, difficulty in verifying model accuracy, strong subjectivity in transforming classical strategies; Directions: Neural-symbolic fusion, multi-cultural strategy library, human-machine hybrid systems, ethical norm research.

This project connects Eastern wisdom with cutting-edge AI technology, promoting AI from isolated problem-solving to improving social interaction capabilities. True intelligence lies not only in performance but also in establishing meaningful connections with others—this is the core value of the interactive cognition framework.