LLM alignment techniques currently struggle with enforcing desired characteristics and harmlessness of outputs over long conversational contexts and chains-of-thought. In this paper we present the scaling problem, a mathematical formulation of this difficulty, and propose interruptions as a means to achieve LLM alignment in scaling contexts. We call this reinforcement in context. Paper structure is as follows: section 1 is this introduction and section 2 presents the scaling problem. In section 3 we describe interruptions as a means to solve the alignment scaling problem. In section 4 we discuss consequences and limitations and in section 5 we highlight avenues for future research.

In high-stakes scenarios-such as self-harm, legal, or medical queries-LLMs must be both trustworthy and helpful. However, these goals often conflict. We propose priority alignment, a new alignment paradigm that enforces a strict "trustworthy-before-helpful" ordering: optimization of helpfulness is conditioned on first meeting trustworthy thresholds (e.g., harmlessness or honesty). To realize this, we introduce Self-Priority Alignment (SPA)-a fully unsupervised framework that generates diverse responses, self-evaluates them and refines them by the model itself, and applies dual-criterion denoising to remove inconsistency and control variance. From this, SPA constructs lexicographically ordered preference pairs and fine-tunes the model using an uncertainty-weighted alignment loss that emphasizes high-confidence, high-gap decisions. Experiments across multiple benchmarks show that SPA improves helpfulness without compromising safety, outperforming strong baselines while preserving general capabilities. Our results demonstrate that SPA provides a scalable and interpretable alignment strategy for critical LLM applications.

Adopting Large language models (LLMs) in organizations potentially revolutionizes our lives and work. However, they can generate off-topic, discriminating, or harmful content. This AI alignment problem often stems from misspecifications during the LLM adoption, unnoticed by the principal due to the LLM's black-box nature. While various research disciplines investigated AI alignment, they neither address the information asymmetries between organizational adopters and black-box LLM agents nor consider organizational AI adoption processes. Therefore, we propose LLM ATLAS (LLM Agency Theory-Led Alignment Strategy) a conceptual framework grounded in agency (contract) theory, to mitigate alignment problems during organizational LLM adoption. We conduct a conceptual literature analysis using the organizational LLM adoption phases and the agency theory as concepts. Our approach results in (1) providing an extended literature analysis process specific to AI alignment methods during organizational LLM adoption and (2) providing a first LLM alignment problem-solution space.

Reward design plays a pivotal role in aligning large language models (LLMs) with human values, serving as the bridge between feedback signals and model optimization. This survey provides a structured organization of reward modeling and addresses three key aspects: mathematical formulation, construction practices, and interaction with optimization paradigms. Building on this, it develops a macro-level taxonomy that characterizes reward mechanisms along complementary dimensions, thereby offering both conceptual clarity and practical guidance for alignment research. The progression of LLM alignment can be understood as a continuous refinement of reward design strategies, with recent developments highlighting paradigm shifts from reinforcement learning (RL)-based to RL-free optimization and from single-task to multi-objective and complex settings.

Human preferences are widely used to align large language models (LLMs) through methods such as reinforcement learning from human feedback (RLHF). However, the current user interfaces require annotators to compare text paragraphs, which is cognitively challenging when the texts are long or unfamiliar. This paper contributes by studying the decomposition principle as an approach to improving the quality of human feedback for LLM alignment. This approach breaks down the text into individual claims instead of directly comparing two long-form text responses. Based on the principle, we build a novel user interface DxHF. It enhances the comparison process by showing decomposed claims, visually encoding the relevance of claims to the conversation and linking similar claims. This allows users to skim through key information and identify differences for better and quicker judgment. Our technical evaluation shows evidence that decomposition generally improves feedback accuracy regarding the ground truth, particularly for users with uncertainty. A crowdsourcing study with 160 participants indicates that using DxHF improves feedback accuracy by an average of 5%, although it increases the average feedback time by 18 seconds. Notably, accuracy is significantly higher in situations where users have less certainty. The finding of the study highlights the potential of HCI as an effective method for improving human-AI alignment.

Human preference data plays a critical role in aligning large language models (LLMs) with human values. However, collecting such data is often expensive and inefficient, posing a significant scalability challenge. To address this, we introduce Alignment Data Map, a GPT-4o-assisted tool for analyzing and diagnosing preference data. Using GPT-4o as a proxy for LLM alignment, we compute alignment scores for LLM-generated responses to instructions from existing preference datasets. These scores are then used to construct an Alignment Data Map based on their mean and variance. Our experiments show that using only 33 percent of the data, specifically samples in the high-mean, low-variance region, achieves performance comparable to or better than using the entire dataset. This finding suggests that the Alignment Data Map can significantly improve data collection efficiency by identifying high-quality samples for LLM alignment without requiring explicit annotations. Moreover, the Alignment Data Map can diagnose existing preference datasets. Our analysis shows that it effectively detects low-impact or potentially misannotated samples. Source code is available online.

As large language models (LLMs) become more powerful and pervasive across society, ensuring these systems are beneficial, safe, and aligned with human values is crucial. Current alignment techniques, like Constitutional AI (CAI), involve complex iterative processes. This paper argues that the Method of Wide Reflective Equilibrium (MWRE) -- a well-established coherentist moral methodology -- offers a uniquely apt framework for understanding current LLM alignment efforts. Moreover, this methodology can substantively augment these processes by providing concrete pathways for improving their dynamic revisability, procedural legitimacy, and overall ethical grounding. Together, these enhancements can help produce more robust and ethically defensible outcomes. MWRE, emphasizing the achievement of coherence between our considered moral judgments, guiding moral principles, and relevant background theories, arguably better represents the intricate reality of LLM alignment and offers a more robust path to justification than prevailing foundationalist models or simplistic input-output evaluations. While current methods like CAI bear a structural resemblance to MWRE, they often lack its crucial emphasis on dynamic, bi-directional revision of principles and the procedural legitimacy derived from such a process. While acknowledging various disanalogies (e.g., consciousness, genuine understanding in LLMs), the paper demonstrates that MWRE serves as a valuable heuristic for critically analyzing current alignment efforts and for guiding the future development of more ethically sound and justifiably aligned AI systems.

Aligning human preference and value is an important requirement for contemporary foundation models. State-of-the-art techniques such as Reinforcement Learning from Human Feedback (RLHF) often consist of two stages: 1) supervised fine-tuning (SFT), where the model is fine-tuned by learning from human demonstration data; 2) Preference learning, where preference data is used to learn a reward model, which is in turn used by a reinforcement learning (RL) step to fine-tune the model. Such reward model serves as a proxy to human preference, and it is critical to guide the RL step towards improving the model quality. In this work, we argue that the SFT stage significantly benefits from learning a reward model as well. Instead of using the human demonstration data directly via supervised learning, we propose to leverage an Inverse Reinforcement Learning (IRL) technique to {\it simultaneously} build an reward model and a policy model.

Aligning foundation models is essential for their safe and trustworthy deployment. However, traditional fine-tuning methods are computationally intensive and require updating billions of model parameters. A promising alternative, alignment via decoding, adjusts the response distribution directly without model updates to maximize a target reward r, thus providing a lightweight and adaptable framework for alignment. However, principled decoding methods rely on oracle access to an optimal Q-function ( Q *), which is often unavailable in practice. Hence, prior SoTA methods either approximate this Q * using Q {\pi_{\text{sft}}} (derived from the reference \texttt{SFT} model) or rely on short-term rewards, resulting in sub-optimal decoding performance. In this work, we propose \texttt{Transfer Q} *, which implicitly estimates the optimal value function for a target reward r through a baseline model \rho_{\texttt{BL}} aligned with a baseline reward r_{\texttt{BL}} (which can be different from the target reward r).

The recent success of using human preferences to align large language models (LLMs) has significantly improved their performance in various downstream tasks like question answering, mathematical reasoning, and code generation. However,3 achieving effective LLM alignment depends on high-quality human preference datasets. Collecting these datasets requires human preference annotation, which is costly and resource-intensive, necessitating efficient active data selection methods. Existing methods either lack a strong theoretical foundation or depend on restrictive reward function assumptions (e.g., linearity). To this end, we propose an algorithm, ActiveDPO, that uses a theoretically grounded data selection criterion for non-linear reward functions while directly leveraging the LLM itself to parameterize the reward model that is used for active data selection. As a result, ActiveDPO explicitly accounts for the influence of LLM on data selection, unlike methods that select the data without considering the LLM that is being aligned, thereby leading to more effective and efficient data collection. Extensive experiments show that ActiveDPO outperforms existing methods across various models and datasets.