Guardado en:
Detalles Bibliográficos
Autor principal: Qiang, Zou
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:https://arxiv.org/abs/2603.19182
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908901522276352
author Qiang, Zou
author_facet Qiang, Zou
contents Large language models (LLMs) demonstrate strong generative capabilities but remain vulnerable to hallucination and unreliable reasoning under adversarial prompting. Existing safety approaches -- such as reinforcement learning from human feedback (RLHF) and output filtering -- primarily operate at the behavioral level and may lack explicit architectural mechanisms for enforcing reasoning process integrity. This paper proposes the Box Maze framework, a conceptual process-control architecture that decomposes LLM reasoning into three explicit layers: memory grounding, structured inference, and boundary enforcement. We introduce preliminary simulation-based evaluation involving progressive boundary erosion scenarios across multiple heterogeneous LLM systems (DeepSeek-V3, Doubao, Qwen). Results from n=50 adversarial scenarios suggest that explicit cognitive control layers may improve consistency in boundary maintenance, with architectural constraints reducing boundary failure rates from approximately 40% (baseline RLHF) to below 1% under adversarial conditions. While current validation is simulation-based, these preliminary results indicate that process-level control may offer a promising direction for improving reliability in large language model reasoning.
format Preprint
id arxiv_https___arxiv_org_abs_2603_19182
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Box Maze: A Process-Control Architecture for Reliable LLM Reasoning
Qiang, Zou
Artificial Intelligence
Computation and Language
I.2.0
Large language models (LLMs) demonstrate strong generative capabilities but remain vulnerable to hallucination and unreliable reasoning under adversarial prompting. Existing safety approaches -- such as reinforcement learning from human feedback (RLHF) and output filtering -- primarily operate at the behavioral level and may lack explicit architectural mechanisms for enforcing reasoning process integrity. This paper proposes the Box Maze framework, a conceptual process-control architecture that decomposes LLM reasoning into three explicit layers: memory grounding, structured inference, and boundary enforcement. We introduce preliminary simulation-based evaluation involving progressive boundary erosion scenarios across multiple heterogeneous LLM systems (DeepSeek-V3, Doubao, Qwen). Results from n=50 adversarial scenarios suggest that explicit cognitive control layers may improve consistency in boundary maintenance, with architectural constraints reducing boundary failure rates from approximately 40% (baseline RLHF) to below 1% under adversarial conditions. While current validation is simulation-based, these preliminary results indicate that process-level control may offer a promising direction for improving reliability in large language model reasoning.
title Box Maze: A Process-Control Architecture for Reliable LLM Reasoning
topic Artificial Intelligence
Computation and Language
I.2.0
url https://arxiv.org/abs/2603.19182