Homo agenticus in the age of agentic AI: Agency loops, power displacement, and the circulation of responsibility
研究了人类如何将主体性归因于AI系统，导致权力感和责任感转移，形成可管理的主体性循环，对理解人机交互中的权力动态有启发。
As attribution-making creatures, humans constantly seek to locate agency somewhere in their environment. Where we choose to attribute agency has profound consequences for our own sense of power, responsibility, and capacity to act. The rise of agentic AI systems disrupts these attribution processes, as humans encounter technologies that appear to act autonomously and control outcomes affecting human lives. Drawing on psychological research on apparent mental causation, I argue that agency attribution operates as an active process of power redistribution. When humans attribute agency to AI systems, they experience systematic “power displacement”—reduced sense of control and responsibility—even while retaining formal authority. This creates a paradox: the more autonomous AI appears, the less autonomous humans feel, regardless of their actual control. However, this displacement is neither permanent nor unidirectional. Agency circulates through predictable “agency loops”—recursive patterns involving delegation, attribution, contingency, reassertion, and reconfiguration—that can be strategically managed. Humans who understand how agency loops operate and intervene in them through strategic attribution processes can develop new forms of expertise and authority. The future belongs to those who understand that agency is always attributed somewhere—and that where we choose to place it determines who has power to act.