Six Birds Theory Defines Agenthood with Measurable Components

The conceptualization of 'agency' in artificial intelligence has long been a subject of philosophical debate, often conflating persistence with control, making empirical validation challenging. Six Birds Theory (SBT) offers a significant theoretical advancement by providing a type-correct, operationalized account of agenthood. This framework moves beyond abstract discussions to establish measurable components, which is critical for the development of truly controllable and predictable autonomous AI systems, thereby addressing fundamental challenges in AI safety and alignment.

Within SBT, an agent is defined as a maintained theory object whose feasible interface policies can steer external futures while maintaining its own viability. This definition is operationalized through four checkable components: ledger-gated feasibility, a robust viability kernel computed as a greatest fixed point, feasible empowerment (channel capacity) as a proxy for difference-making, and an empirical packaging map whose idempotence defect quantifies objecthood. Experimental validation in a minimal ring-world environment demonstrated key separations, including calibrated null regimes, the collapse of idempotence defect with repair mechanisms, increased empowerment with multi-step protocols, and a monotonic increase in empowerment through operator rewriting.

These findings provide hash-traceable tests that rigorously separate agenthood from mere agency, crucially without making claims about goals, consciousness, or biological organisms. The ability to empirically quantify and test for agentic properties has profound implications for the design, verification, and regulation of advanced AI. This foundational work lays the groundwork for developing more robust and trustworthy AI agents, offering a scientific basis for future AI governance and ensuring that increasingly autonomous systems can be understood and controlled within defined parameters.