Imagen de Portada

Via Negativa for AI Alignment: Why Negative Constraints Are Structurally Superior to Positive Preferences

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autor principal:	Cheng, Quan
Formato:	Preprint
Publicado:	2026
Materias:	Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Semantic Gravity Wells: Why Negative Constraints Backfire
por: Rana, Shailesh
Publicado: (2026)

Beyond Preferences in AI Alignment
por: Zhi-Xuan, Tan, et al.
Publicado: (2024)

Why the Valuable Capabilities of LLMs Are Precisely the Unexplainable Ones
por: Cheng, Quan
Publicado: (2026)

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

Why AI Alignment Failure Is Structural: Learned Human Interaction Structures and AGI as an Endogenous Evolutionary Shock
por: Sornette, Didier, et al.
Publicado: (2026)

Alignment Revisited: Are Large Language Models Consistent in Stated and Revealed Preferences?
por: Gu, Zhuojun, et al.
Publicado: (2025)

On Diversified Preferences of Large Language Model Alignment
por: Zeng, Dun, et al.
Publicado: (2023)

Preference Learning for AI Alignment: a Causal Perspective
por: Kobalczyk, Katarzyna, et al.
Publicado: (2025)

HPS: Hard Preference Sampling for Human Preference Alignment
por: Zou, Xiandong, et al.
Publicado: (2025)

FlipGuard: Defending Preference Alignment against Update Regression with Constrained Optimization
por: Zhu, Mingye, et al.
Publicado: (2024)

Strong Preferences Affect the Robustness of Preference Models and Value Alignment
por: Xu, Ziwei, et al.
Publicado: (2024)

On Negative-aware Preference Optimization for Recommendation
por: Ding, Chenlu, et al.
Publicado: (2025)

Negating Negatives: Alignment with Human Negative Samples via Distributional Dispreference Optimization
por: Duan, Shitong, et al.
Publicado: (2024)

Unified Preference Optimization: Language Model Alignment Beyond the Preference Frontier
por: Badrinath, Anirudhan, et al.
Publicado: (2024)

Alignment Imprint: Zero-Shot AI-Generated Text Detection via Provable Preference Discrepancy
por: Wu, Junxi, et al.
Publicado: (2026)

A Revealed Preference Framework for AI Alignment
por: Suleymanov, Elchin
Publicado: (2026)

Position: Safety and Fairness in Agentic AI Depend on Interaction Topology, Not on Model Scale or Alignment
por: Bajaj, Tanav Singh, et al.
Publicado: (2026)

Adaptive Helpfulness-Harmlessness Alignment with Preference Vectors
por: Liang, Ren-Wei, et al.
Publicado: (2025)

Implicit Safety Alignment from Crowd Preferences
por: Lin, Qian, et al.
Publicado: (2026)

Architectural Constraints Alignment in AI-assisted, Platform-based Service Development
por: Irion, Julius, et al.
Publicado: (2026)

Position: Towards Bidirectional Human-AI Alignment
por: Shen, Hua, et al.
Publicado: (2024)

Direct Alignment with Heterogeneous Preferences
por: Shirali, Ali, et al.
Publicado: (2025)

Why AI Safety Requires Uncertainty, Incomplete Preferences, and Non-Archimedean Utilities
por: Benavoli, Alessio, et al.
Publicado: (2025)

Out of Control -- Why Alignment Needs Formal Control Theory (and an Alignment Control Stack)
por: Perrier, Elija
Publicado: (2025)

ConstraintBench: Benchmarking LLM Constraint Reasoning on Direct Optimization
por: Tso, Joseph, et al.
Publicado: (2026)

Implicit Preference Alignment for Human Image Animation
por: Wang, Yuanzhi, et al.
Publicado: (2026)

Chain-of-Thought as a Lens: Evaluating Structured Reasoning Alignment between Human Preferences and Large Language Models
por: Wang, Boxuan, et al.
Publicado: (2025)

Word Alignment as Preference for Machine Translation
por: Wu, Qiyu, et al.
Publicado: (2024)

Preference Ranking Optimization for Human Alignment
por: Song, Feifan, et al.
Publicado: (2023)

Disentangling Preference Representation and Text Generation for Efficient Individual Preference Alignment
por: Zhang, Jianfei, et al.
Publicado: (2024)

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

Adaptive Alignment: Dynamic Preference Adjustments via Multi-Objective Reinforcement Learning for Pluralistic AI
por: Harland, Hadassah, et al.
Publicado: (2024)

Correcting Negative Bias in Large Language Models through Negative Attention Score Alignment
por: Yu, Sangwon, et al.
Publicado: (2024)

ClinAlign: Scaling Healthcare Alignment from Clinician Preference
por: Lyu, Shiwei, et al.
Publicado: (2026)

Improving Safety Alignment via Balanced Direct Preference Optimization
por: Zhao, Shiji, et al.
Publicado: (2026)

The Anatomy of Alignment: Decomposing Preference Optimization by Steering Sparse Features
por: Ferrao, Jeremias, et al.
Publicado: (2025)

Graph Unlearning Meets Influence-aware Negative Preference Optimization
por: Chen, Qiang, et al.
Publicado: (2025)

Balancing Safety and Helpfulness in Healthcare AI Assistants through Iterative Preference Alignment
por: Nghiem, Huy, et al.
Publicado: (2025)

Negative as Positive: Enhancing Out-of-distribution Generalization for Graph Contrastive Learning
por: Wang, Zixu, et al.
Publicado: (2024)

Adversarial Preference Learning for Robust LLM Alignment
por: Wang, Yuanfu, et al.
Publicado: (2025)