Beyond Surface Alignment: Grounding the Dynamics of Situational Understanding and Generative Control in LLMs

• URL: http://arxiv.org/abs/2608.29610v1
• Date: Sun, 30 Aug 2026 07:04:03 GMT
• Title: Beyond Surface Alignment: Grounding the Dynamics of Situational Understanding and Generative Control in LLMs
• Abstract summary: The current alignment tuning paradigm for Large Language Models (LLMs) prioritizes surface-level behaviors.<n>This thesis argues that such surface alignment masks a lack of grounding, creating models that are stylistically confident but situationally brittle.
• Score: 4.531912685912965
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: The current alignment tuning paradigm for Large Language Models (LLMs) prioritizes surface-level behaviors -- fluency, safety, and tonal consistency. While effective for casual chat, this thesis argues that such surface alignment masks a lack of grounding, creating models that are stylistically confident but situationally brittle. We propose a framework of Grounded Alignment, analyzing how models process context (Input) and structure generation (Output), then aligning these grounded behaviors to human needs. First, we evaluate failures in Situational Grounding. SitTest shows that despite large context windows, state-of-the-art models struggle to maintain a consistent "mental model" of a changing environment. ReCode further shows that models rely on surface heuristics rather than deep syntactic dependencies: they "read" extensive histories without truly "understanding" the evolving situation. Second, we evaluate Generative Grounding. We introduce the Branching Factor (BF) to map LLM generation, finding that standard alignment tuning constricts this landscape into premature stylistic collapse. Hindsight further shows that models often fail to understand their own generations. Finally, we propose Dynamic Control for grounded interaction. AI Realtor demonstrates context engineering to compensate for poor situational grounding. Base-Aligned Model Collaboration decouples exploration from stylistic constraints. We also present Annealed Sampling for verifiable reinforcement learning and apply these ideas to Addiction Support, where model-generated rationalization offers a communication interface for high-stakes domains. Collectively, this work moves beyond surface alignment toward agents anchored in both context and generation.

Related papers

• Towards Predictive, Aligned, and Scalable Robot Learning [11.830982964522605]
  We introduce Lumo-2, a latent world-action model that generates actions by reasoning over world dynamics in latent space.<n>We study latent world modelling and modality alignment, analyzing their roles in scaling laws and out-of-distribution generalization.<n>Results show that Lumo-2 consistently outperforms strong vision-language-action (VLA) and world-action model (WAM) baselines.
  arXiv  Detail & Related papers  (2026-07-13T08:53:34Z)
• The Art of Interrogation: Consistency Amplifies Factuality in Spatial Reasoning [65.67266333751569]
  Current Large Reasoning Models (LRMs) exhibit remarkable general capabilities but significantly underperform in spatial reasoning tasks.<n>We propose a self-supervised reinforcement learning framework that targets the internal reasoning process without requiring ground-truth annotations.<n>We show that this label-free consistency training approaches the accuracy of models trained with ground-truth supervision and achieves similar generalization across diverse tasks and data domains.
  arXiv  Detail & Related papers  (2026-06-10T10:50:06Z)
• iVGR: Internalizing Visually Grounded Reasoning for MLLMs with Reinforcement Learning [35.83366559347647]
  Visually grounded Chain-of-Thought (CoT) has emerged as a promising paradigm to enhance fine-grained perception in large language models (MLLMs)<n>Mandating explicit object boxes in visually grounded CoT during inference often degrades performance compared to standard textual CoT.<n>We propose Internalizing Visually Grounded Reasoning (textbfiVGR), a novel reinforcement learning framework that transfers localization capabilities into the textual reasoning process.
  arXiv  Detail & Related papers  (2026-05-29T10:07:25Z)
• Understanding Degradation with Vision Language Model [56.09241449206817]
  Understanding visual degradations is a critical yet challenging problem in computer vision.<n>We introduce DU-VLM, a multimodal chain-of-thought model trained with supervised fine-tuning and reinforcement learning.<n>We also introduce textbfDU-110k, a large-scale dataset comprising 110,000 clean-degraded pairs with grounded physical annotations.
  arXiv  Detail & Related papers  (2026-02-04T13:51:15Z)
• Dynamics-Aligned Latent Imagination in Contextual World Models for Zero-Shot Generalization [1.6332728502735252]
  We introduce Dynamics-Aligned Latent Imagination (DALI), a framework that infers latent context representations from agent-environment interactions.<n>DALI generates actionable representations conditioning the world model and policy, bridging perception and control.<n>On challenging cMDP benchmarks, DALI achieves significant gains over context-unaware baselines.
  arXiv  Detail & Related papers  (2025-08-27T22:02:56Z)
• Context-Informed Grounding Supervision [102.11698329887226]
  Context-INformed Grounding Supervision (CINGS) is a post-training supervision in which the model is trained with relevant context prepended to the response.<n>Our experiments demonstrate that models trained with CINGS exhibit stronger grounding in both textual and visual domains.
  arXiv  Detail & Related papers  (2025-06-18T14:13:56Z)
• Foundational Models Defining a New Era in Vision: A Survey and Outlook [151.49434496615427]
  Vision systems to see and reason about the compositional nature of visual scenes are fundamental to understanding our world. The models learned to bridge the gap between such modalities coupled with large-scale training data facilitate contextual reasoning, generalization, and prompt capabilities at test time. The output of such models can be modified through human-provided prompts without retraining, e.g., segmenting a particular object by providing a bounding box, having interactive dialogues by asking questions about an image or video scene or manipulating the robot's behavior through language instructions.
  arXiv  Detail & Related papers  (2023-07-25T17:59:18Z)
• Grounded Decoding: Guiding Text Generation with Grounded Models for Embodied Agents [111.15288256221764]
  Grounded-decoding project aims to solve complex, long-horizon tasks in a robotic setting by leveraging the knowledge of both models. We frame this as a problem similar to probabilistic filtering: decode a sequence that both has high probability under the language model and high probability under a set of grounded model objectives. We demonstrate how such grounded models can be obtained across three simulation and real-world domains, and that the proposed decoding strategy is able to solve complex, long-horizon tasks in a robotic setting by leveraging the knowledge of both models.
  arXiv  Detail & Related papers  (2023-03-01T22:58:50Z)
• Large Language Models with Controllable Working Memory [64.71038763708161]
  Large language models (LLMs) have led to a series of breakthroughs in natural language processing (NLP) What further sets these models apart is the massive amounts of world knowledge they internalize during pretraining. How the model's world knowledge interacts with the factual information presented in the context remains under explored.
  arXiv  Detail & Related papers  (2022-11-09T18:58:29Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.