Imagen de Portada

Out of Control -- Why Alignment Needs Formal Control Theory (and an Alignment Control Stack)

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autor principal:	Perrier, Elija
Formato:	Preprint
Publicado:	2025
Materias:	Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Operationalising Extended Cognition: Formal Metrics for Corporate Knowledge and Legal Accountability
por: Perrier, Elija
Publicado: (2025)

Hamiltonian Formalism for Comparing Quantum and Classical Intelligence
por: Perrier, Elija
Publicado: (2025)

Towards Measurement Theory for Artificial Intelligence
por: Perrier, Elija
Publicado: (2025)

Typed Chain-of-Thought: A Curry-Howard Framework for Verifying LLM Reasoning
por: Perrier, Elija
Publicado: (2025)

Statistical Scenario Modelling and Lookalike Distributions for Multi-Variate AI Risk
por: Perrier, Elija
Publicado: (2025)

Deconstructing Superintelligence: Identity, Self-Modification and Différance
por: Perrier, Elija
Publicado: (2026)

K-P Quantum Neural Networks
por: Perrier, Elija
Publicado: (2025)

Quantum AIXI: Universal Intelligence via Quantum Information
por: Perrier, Elija
Publicado: (2025)

Threshold Crossings as Tail Events for Catastrophic AI Risk
por: Perrier, Elija
Publicado: (2025)

Watts-per-Intelligence Part II: Algorithmic Catalysis
por: Perrier, Elija
Publicado: (2026)

Post-AGI Economies: Autonomy and the First Fundamental Theorem of Welfare Economics
por: Perrier, Elija
Publicado: (2026)

Time, Identity and Consciousness in Language Model Agents
por: Perrier, Elija, et al.
Publicado: (2026)

Quantum AGI: Ontological Foundations
por: Perrier, Elija, et al.
Publicado: (2025)

Position: Stop Acting Like Language Model Agents Are Normal Agents
por: Perrier, Elija, et al.
Publicado: (2025)

Agent Identity Evals: Measuring Agentic Identity
por: Perrier, Elija, et al.
Publicado: (2025)

Beyond Ordinal Preferences: Why Alignment Needs Cardinal Human Feedback
por: Whitfill, Parker, et al.
Publicado: (2025)

MOSAIC: Composable Safety Alignment with Modular Control Tokens
por: Peng, Jingyu, et al.
Publicado: (2026)

Controllable Preference Optimization: Toward Controllable Multi-Objective Alignment
por: Guo, Yiju, et al.
Publicado: (2024)

CAP: Controllable Alignment Prompting for Unlearning in LLMs
por: Wang, Zhaokun, et al.
Publicado: (2026)

CBF-LLM: Safe Control for LLM Alignment
por: Miyaoka, Yuya, et al.
Publicado: (2024)

Collab: Controlled Decoding using Mixture of Agents for LLM Alignment
por: Chakraborty, Souradip, et al.
Publicado: (2025)

Safe Transformer: An Explicit Safety Bit For Interpretable And Controllable Alignment
por: Feng, Jingyuan, et al.
Publicado: (2026)

Universe Routing: Why Self-Evolving Agents Need Epistemic Control
por: Wang, Zhaohui Geoffrey
Publicado: (2026)

Watts-Per-Intelligence: Part I (Energy Efficiency)
por: Perrier, Elija
Publicado: (2025)

Quantum Geometric Machine Learning
por: Perrier, Elija
Publicado: (2024)

Position: Assistive Agents Need Accessibility Alignment
por: Hu, Jie, et al.
Publicado: (2026)

Inference-Time Alignment Control for Diffusion Models with Reinforcement Learning Guidance
por: Jin, Luozhijie, et al.
Publicado: (2025)

Position: Capability Control Should be a Separate Goal From Alignment
por: Siddiqui, Shoaib Ahmed, et al.
Publicado: (2026)

SudoLM: Learning Access Control of Parametric Knowledge with Authorization Alignment
por: Liu, Qin, et al.
Publicado: (2024)

Controllable Safety Alignment: Inference-Time Adaptation to Diverse Safety Requirements
por: Zhang, Jingyu, et al.
Publicado: (2024)

Analogous Alignments: Digital "Formally" meets Analog
por: Mohanty, Hansa, et al.
Publicado: (2024)

GAC: Stabilizing Asynchronous RL Training for LLMs via Gradient Alignment Control
por: Xu, Haofeng, et al.
Publicado: (2026)

C-MORAL: Controllable Multi-Objective Molecular Optimization with Reinforcement Alignment for LLMs
por: Gao, Rui, et al.
Publicado: (2026)

The Geometry of Compromise: Unlocking Generative Capabilities via Controllable Modality Alignment
por: Liu, Hongyuan, et al.
Publicado: (2026)

Expected Value Alignment for Generative Reward Modeling in Formal Mathematics Verification
por: Ji, Shihao, et al.
Publicado: (2026)

JAM: A Tiny Flow-based Song Generator with Fine-grained Controllability and Aesthetic Alignment
por: Liu, Renhang, et al.
Publicado: (2025)

Maximize Your Diffusion: A Study into Reward Maximization and Alignment for Diffusion-based Control
por: Huh, Dom, et al.
Publicado: (2025)

Medoid Prototype Alignment for Cross-Plant Unknown Attack Detection in Industrial Control Systems
por: Wang, Luyao
Publicado: (2026)

Infrastructure for AI Agents
por: Chan, Alan, et al.
Publicado: (2025)

How Alignment Routes: Localizing, Scaling, and Controlling Policy Circuits in Language Models
por: Frank, Gregory N.
Publicado: (2026)