cs.AIOct 12, 2026

Agency as a New Organizing Abstraction for Computer Science

Authors: Ian Foster

Organizations: Argonne National Laboratory and The University of Chicago

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Advances in large-scale machine learning have enabled agentic AI systems: persistent, autonomous computational entities that pursue goals, initiate actions, and interact with open environments. These systems are often framed as applications of foundation models. We argue that this framing is incomplete: agency is also a candidate new organizing abstraction for computer science. What is new is not the concept of an agent, nor runtime policy generation per se, but its general-purpose, open-ended form, in which systems can synthesize courses of action from goals, observations, and memory, over tools and environments for which no complete task-specific model was specified in advance. Thus agency becomes a systems-level abstraction as well as an AI modeling concept. Where the process abstraction organizes computation around executing programs, agency organizes it around pursuing goals under bounded resources: a process executes a program; an agent pursues a goal. From this perspective, the recurring failures of deployed agentic systems (runaway resource consumption, behavior redirected by adversarial content, opaque continuous operation) are symptoms of structural abstraction gaps, not isolated engineering defects. We examine these gaps across six dimensions (execution, interfaces, correctness, oversight, coordination, and persistence) and argue that closing them requires coordinated advances in systems,

Figures & tables

Figure 1: From models to agents. A foundation model is a stateless inference engine that maps prompts to responses; an agent embeds such a model in a persistent perceive–reason–act loop, accumulating state and interacting continuously with an open environment.
DimensionConventional organizing focusAgency makes first-class
ExecutionResource limitsResource budgets
InterfacesProtocolsContracts
CorrectnessFunctional correctnessBehavioral correctness
OversightCommandsAttention allocation
CoordinationSchedulingNegotiation
PersistenceStateMemory
Table 1: Six dimensions along which agency extends existing abstractions. In each, prior work addresses the right-hand concern in specialized settings; agency makes it a general and intrinsic requirement (Section 5 ).
Figure 2: Organizing abstractions address distinct concerns and are complements rather than successors. We ask whether autonomous goal pursuit is a concern of comparable generality, and the agent its organizing abstraction.
Figure 3: Four properties characterize agentic operation. Each has precedent in earlier systems; what is new is their combination in a single general-purpose system, enabled by foundation models. The deeper distinction lies in how behavior is produced (Section 3.2 ).
Figure 4: Architecture of an agentic system. The agent comprises ephemeral working state, goals ( GG ), persistent memory ( MM ), and a policy generator ( Π\Pi ). A foundation model provides general-purpose reasoning for policy generation. The agent interacts with an open environment ( EE ) through actions and observations, running a persistent control loop under human oversight.
Figure 5: A runaway-agent scenario. An agent monitoring a cloud deployment detects an anomaly and responds autonomously; the cost of its actions becomes visible only after budgets have been exceeded. Reactive resource control arrives too late; agentic systems require proactive constraint enforcement.
Figure 6: Lifecycle mismatch. Batch jobs run toward termination; services are organized around periods of request handling; agents blur these boundaries: intermittently active, spawning children that outlive their parent, and accumulating context that makes termination costly.
Figure 7: Failure attribution. Traditional debugging seeks a single faulty frame in a linear execution trace (left); agentic failures emerge from partial orders of locally reasonable decisions that compound into a globally undesirable outcome (right).
Observed Failure ModeAbstraction GapResulting Research Challenges
Autonomous resource consumption and unbounded action initiationExecution models assume externally invoked computation with reactive resource controlProactive resource budgeting, anticipatory constraint enforcement, isolation for self-directed processes
Ambiguous agent lifecycles and costly terminationProcess and service abstractions assume clear start, run, and termination phasesLifecycle models for persistent agents, state preservation and transfer, principled shutdown semantics
Distributed failure attribution across long decision sequencesDebugging and verification assume localized faults and linear executionCausal tracing across reasoning-action loops, partial-order explanations, long-horizon accountability
Misplaced responsibility at system boundariesControl and correctness are assumed to reside outside the computational componentNew interface contracts, responsibility assignment, end-to-end reasoning for autonomous initiators
Behavior redirected by adversarial contentAbstractions assume a separation between trusted code and untrusted dataTrust-boundary tracking, provenance-aware interpretation, limits on untrusted influence over actions
Opaque behavior under continuous operationInteraction models assume explicit user commands and episodic executionOversight interfaces, escalation policies, scalable human supervision mechanisms
Table 2: Characteristic failure modes of agentic systems reflect mismatches between persistent, autonomous behavior and existing computational abstractions.
Figure 8: Oversight at scale. A single supervisor cannot review every decision of a fleet of continuously operating agents; oversight becomes a problem of allocating scarce human attention, with escalation as a first-class mechanism.
Figure 9: The agent abstraction in synthesis. The agent of Figure 4 , now exercising authority delegated by its principal, embedded in a governing substrate whose mechanisms (budgets, contracts, provenance, runtime monitors, oversight) make generated behavior governable.
More in this topic

Agentic AI

Agentic artificial intelligence studies systems that plan, use tools, learn from feedback, and act toward goals with varying degrees of autonomy.

See all 212 papers

Explore similar work

CardsList
Jun 22, 2026

Critique of Agent Model

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 *internalized within the system itself* rather than assembled through external scaffolding. This distinction between *agentic* systems, whose competence resides in engineered workflows, and *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.

Eric Xing, Mingkai Deng, Jinyu Hou

Institute of Foundation Models, Mohamed bin Zayed University of Artificial Intelligence · School of Computer Science, Carnegie Mellon University

AI Agent SafetyAI Agent ArchitectureAgentic AI
AskPDF
May 25, 2026

From Model Scaling to System Scaling: Scaling the Harness in Agentic AI

This paper studies the next major bottleneck in agentic AI as system scaling, not only model scaling: the design of auditable, persistent, modular, and verifiable architectures around foundation models. We refer to this shift as scaling the harness: treating the structured execution layer around a foundation model as a first-class object of design, evaluation, and optimization. Although recent large language models enable agents to use tools, retrieve information, maintain memory, and execute long-horizon workflows, evaluation remains largely model-centric, often reducing agents to final-task success while treating memory, retrieval, tool use, orchestration, verification, and governance as secondary implementation details. This framing is increasingly inadequate because agent performance emerges from the interaction among the foundation model, memory substrate, context constructor, skill-routing layer, orchestration loop, and verification-and-governance layer. Together, these components form the agent harness, which translates model capability into long-horizon agent behavior. We study scaling the harness through three core bottlenecks: context governance, trustworthy memory, and dynamic skill routing, together with the orchestration and governance mechanisms that coordinate and constrain them. We further outline a research agenda for harness-level benchmarks that go beyond one-shot task success to measure trajectory quality, memory hygiene, context efficiency, communication fidelity, verification cost, and safe evolution over time. To make the discussion concrete, we develop CheetahClaws: https://github.com/SafeRL-Lab/cheetahclaws, a Python-native reference harness, and compare it with Claude Code and OpenClaw. Our main claim is that future progress in agentic AI will depend as much on system design as on stronger foundation models.

Shangding Gu

UC Berkeley

LLM Agent OrchestrationAI Agent BenchmarksAI Agent GovernanceMemory-Augmented AgentsAgentic AI
AskPDF
May 13, 2026

Position: Agentic AI System Is a Foreseeable Pathway to AGI

Is monolithic scaling the only path to AGI? This paper challenges the dogma that purely scaling a single model is sufficient to achieve Artificial General Intelligence. Instead, we identify Agentic AI as a necessary paradigm for mastering the complex, heterogeneous distribution of real-world tasks. Through rigorous theoretical derivations, we contrast the optimization constraints of monolithic learners against the efficiency of Agentic systems, progressing from simple routing mechanisms to general Directed Acyclic Graph (DAG) topologies. We demonstrate that Agentic AI achieves exponentially superior generalization and sample efficiency. Finally, we discuss the connection to Mixture-of-Experts, reinterpret the instability of current multi-agent frameworks, and call for greater research focus on Agentic AI.

Junwei Liao, Shuai Li, Muning Wen +2

1Shanghai Jiao Tong University 2Shanghai Innovation Institute · 1Shanghai Jiao Tong University · University College London.

Multi-Agent SystemsAgentic AIArtificial General Intelligence
AskPDF