Critique of Agent Model

2026-06-22

AI Summary

This work addresses the conflation of large language models with genuine agents, a terminological imprecision that obscures the distinction between automated systems and entities possessing intrinsic autonomy, thereby risking capability overestimation and safety hazards. Drawing on Cartesian notions of autonomy and science-fictional conceptions of self-directed existence, the paper delineates extrinsically orchestrated “agentic” systems from intrinsically autonomous “agentive” systems along five dimensions: goals, identity, decision-making, self-regulation, and learning. It introduces the novel Goal–Identity–Configurator (GIC) architecture, which asserts that agent capabilities must be internalized rather than externally scripted. By integrating hierarchical goal decomposition, evolving identity, simulation-based reasoning driven by an independent world model, and reality-grounded autonomous learning, the framework establishes a theoretical foundation for general-purpose agents capable of operating in open-world environments, offering a new paradigm for auditability, controllability, and safety in highly autonomous systems.

Abstract

What is an agent? What constitutes agency? With the rise of Large Language Model (LLM) systems marketed as ``coding agents'', ``AI co-scientists'', and other ``agentic" tools that promise to drive up productivity, and at the same time, ``existential" concerns such as AI escaping human control with destructive power under a speculative ``machine agency" against humans, it has become essential to clarify where automation ends and agency begins, both for building capable systems and for understanding whether and what to fear. Drawing on Descartes' grounding of agency in independent thought, and on portrayals of autonomous beings in science fiction, we survey the current landscape of AI agents, and analyze agent architectures along five dimensions: goal, identity, decision-making, self-regulation, and learning. Specifically, we argue that genuine agency requires these structures to be \emph{internalized within the system itself} rather than assembled through external scaffolding. This distinction between \emph{agentic} systems, whose competence resides in engineered workflows, and \emph{agentive} systems, whose capabilities (including social interaction) arise endogenously, defines the boundary between systems designed for prescribed tasks, and those capable of operating in the open world with true autonomy. Building on this analysis, we propose the Goal-Identity-Configurator (GIC) architecture for a general-purpose agent model, combining hierarchical goal decomposition, identity evolution, simulative reasoning grounded in a separately trained world model, learned self-regulation, and self-directed learning from both real and simulated experience. Furthermore, we share insight on the auditability, controllability, and safety of agentive systems that possess greater autonomy and ``agency", but remain under human oversight.

Problem

• agency
• autonomy
• AI agents
• machine agency
• agent model

Innovation

• agentive systems
• Goal-Identity-Configurator
• internalized agency
• simulative reasoning
• autonomous learning