Understanding the Process of Human-AI Value Alignment

Background: Value alignment in computer science research is often used to refer to the process of aligning artificial intelligence with humans, but the way the phrase is used often lacks precision. Objectives: In this paper, we conduct a systematic literature review to advance the understanding of value alignment in artificial intelligence by characterising the topic in the context of its research literature. We use this to suggest a more precise definition of the term. Methods: We analyse 172 value alignment research articles that have been published in recent years and synthesise their content using thematic analyses. Results: Our analysis leads to six themes: value alignment drivers & approaches; challenges in value alignment; values in value alignment; cognitive processes in humans and AI; human-agent teaming; and designing and developing value-aligned systems. Conclusions: By analysing these themes in the context of the literature we define value alignment as an ongoing process between humans and autonomous agents that aims to express and implement abstract values in diverse contexts, while managing the cognitive limits of both humans and AI agents and also balancing the conflicting ethical and political demands generated by the values in different groups. Our analysis gives rise to a set of research challenges and opportunities in the field of value alignment for future work.

RepIt: Representing Isolated Targets to Steer Language Models

While activation steering in large language models (LLMs) is a growing area of research, methods can often incur broader effects than desired. This motivates isolation of purer concept vectors to enable targeted interventions and understand LLM behavior at a more granular level. We present RepIt, a simple and data-efficient framework for isolating concept-specific representations. Across five frontier LLMs, RepIt enables precise interventions: it selectively suppresses refusal on targeted concepts while preserving refusal elsewhere, producing models that answer WMD-related questions while still scoring as safe on standard benchmarks. We further show that the corrective signal localizes to just 100-200 neurons and that robust target representations can be extracted from as few as a dozen examples on a single A6000. This efficiency raises a dual concern: manipulations can be performed with modest compute and data to extend to underrepresented data-scarce topics while evading existing benchmarks. By disentangling refusal vectors with RepIt, this work demonstrates that targeted interventions can counteract overgeneralization, laying the foundation for more granular control of model behavior.

Stochastic Streets: A Walk Through Random LLM Address Generation in four European Cities

Large Language Models (LLMs) are capable of solving complex math problems or answer difficult questions on almost any topic, but can they generate random street addresses for European cities?

Don't Change My View: Ideological Bias Auditing in Large Language Models

As large language models (LLMs) become increasingly embedded in products used by millions, their outputs may influence individual beliefs and, cumulatively, shape public opinion. If the behavior of LLMs can be intentionally steered toward specific ideological positions, such as political or religious views, then those who control these systems could gain disproportionate influence over public discourse. Although it remains an open question whether LLMs can reliably be guided toward coherent ideological stances and whether such steering can be effectively prevented, a crucial first step is to develop methods for detecting when such steering attempts occur. In this work, we adapt a previously proposed statistical method to the new context of ideological bias auditing. Our approach carries over the model-agnostic design of the original framework, which does not require access to the internals of the language model. Instead, it identifies potential ideological steering by analyzing distributional shifts in model outputs across prompts that are thematically related to a chosen topic. This design makes the method particularly suitable for auditing proprietary black-box systems. We validate our approach through a series of experiments, demonstrating its practical applicability and its potential to support independent post hoc audits of LLM behavior.

Topic Coverage-based Demonstration Retrieval for In-Context Learning

The effectiveness of in-context learning relies heavily on selecting demonstrations that provide all the necessary information for a given test input. To achieve this, it is crucial to identify and cover fine-grained knowledge requirements. However, prior methods often retrieve demonstrations based solely on embedding similarity or generation probability, resulting in irrelevant or redundant examples. In this paper, we propose TopicK, a topic coverage-based retrieval framework that selects demonstrations to comprehensively cover topic-level knowledge relevant to both the test input and the model. Specifically, TopicK estimates the topics required by the input and assesses the model's knowledge on those topics. TopicK then iteratively selects demonstrations that introduce previously uncovered required topics, in which the model exhibits low topical knowledge. We validate the effectiveness of TopicK through extensive experiments across various datasets and both open- and closed-source LLMs. Our source code is available at https://github.com/WonbinKweon/TopicK_EMNLP2025.

"A 6 or a 9?": Ensemble Learning Through the Multiplicity of Performant Models and Explanations

Creating models from past observations and ensuring their effectiveness on new data is the essence of machine learning. However, selecting models that generalize well remains a challenging task. Related to this topic, the Rashomon Effect refers to cases where multiple models perform similarly well for a given learning problem. This often occurs in real-world scenarios, like the manufacturing process or medical diagnosis, where diverse patterns in data lead to multiple high-performing solutions. We propose the Rashomon Ensemble, a method that strategically selects models from these diverse high-performing solutions to improve generalization. By grouping models based on both their performance and explanations, we construct ensembles that maximize diversity while maintaining predictive accuracy. This selection ensures that each model covers a distinct region of the solution space, making the ensemble more robust to distribution shifts and variations in unseen data. We validate our approach on both open and proprietary collaborative real-world datasets, demonstrating up to 0.20+ AUROC improvements in scenarios where the Rashomon ratio is large. Additionally, we demonstrate tangible benefits for businesses in various real-world applications, highlighting the robustness, practicality, and effectiveness of our approach.

Short-term cognitive fatigue of spatial selective attention after face-to-face conversations in virtual noisy environments

Spatial selective attention is an important asset for communication in cocktail party situations but may be compromised by short-term cognitive fatigue. Here we tested whether an effortful conversation in a highly ecological setting depletes task performance in an auditory spatial selective attention task. Young participants with normal hearing performed the task before and after (1) having a real dyadic face-to-face conversation on a free topic in a virtual reverberant room with simulated interfering conversations and background babble noise at 72 dB SPL for 30 minutes, (2) passively listening to the interfering conversations and babble noise, or (3) having the conversation in quiet. Self-reported perceived effort and fatigue increased after conversations in noise and passive listening relative to the reports after conversations in quiet. In contrast to our expectations, response times in the attention task decreased, rather than increased, after conversation in noise and accuracy did not change systematically in any of the conditions on the group level. Unexpectedly, we observed strong training effects between the individual sessions in our within-subject design even after one hour of training on a different day.

MetaRAG: Metamorphic Testing for Hallucination Detection in RAG Systems

Large Language Models (LLMs) are increasingly deployed in enterprise applications, yet their reliability remains limited by hallucinations, i.e., confident but factually incorrect information. Existing detection approaches, such as SelfCheckGPT and MetaQA, primarily target standalone LLMs and do not address the unique challenges of Retrieval-Augmented Generation (RAG) systems, where responses must be consistent with retrieved evidence. We therefore present MetaRAG, a metamorphic testing framework for hallucination detection in Retrieval-Augmented Generation (RAG) systems. MetaRAG operates in a real-time, unsupervised, black-box setting, requiring neither ground-truth references nor access to model internals, making it suitable for proprietary and high-stakes domains. The framework proceeds in four stages: (1) decompose answers into atomic factoids, (2) generate controlled mutations of each factoid using synonym and antonym substitutions, (3) verify each variant against the retrieved context (synonyms are expected to be entailed and antonyms contradicted), and (4) aggregate penalties for inconsistencies into a response-level hallucination score. Crucially for identity-aware AI, MetaRAG localizes unsupported claims at the factoid span where they occur (e.g., pregnancy-specific precautions, LGBTQ+ refugee rights, or labor eligibility), allowing users to see flagged spans and enabling system designers to configure thresholds and guardrails for identity-sensitive queries. Experiments on a proprietary enterprise dataset illustrate the effectiveness of MetaRAG for detecting hallucinations and enabling trustworthy deployment of RAG-based conversational agents. We also outline a topic-based deployment design that translates MetaRAG's span-level scores into identity-aware safeguards; this design is discussed but not evaluated in our experiments.

Explaining Concept Drift through the Evolution of Group Counterfactuals

Machine learning models in dynamic environments often suffer from concept drift, where changes in the data distribution degrade performance. While detecting this drift is a well-studied topic, explaining how and why the model's decision-making logic changes still remains a significant challenge. In this paper, we introduce a novel methodology to explain concept drift by analyzing the temporal evolution of group-based counterfactual explanations (GCEs). Our approach tracks shifts in the GCEs' cluster centroids and their associated counterfactual action vectors before and after a drift. These evolving GCEs act as an interpretable proxy, revealing structural changes in the model's decision boundary and its underlying rationale. We operationalize this analysis within a three-layer framework that synergistically combines insights from the data layer (distributional shifts), the model layer (prediction disagreement), and our proposed explanation layer. We show that such holistic view allows for a more comprehensive diagnosis of drift, making it possible to distinguish between different root causes, such as a spatial data shift versus a re-labeling of concepts.

A novel method and dataset for depth-guided image deblurring from smartphone Lidar

Modern smartphones are equipped with Lidar sensors providing depth-sensing capabilities. Recent works have shown that this complementary sensor allows to improve various tasks in image processing, including deblurring. However, there is a current lack of datasets with realistic blurred images and paired mobile Lidar depth maps to further study the topic. At the same time, there is also a lack of blind zero-shot methods that can deblur a real image using the depth guidance without requiring extensive training sets of paired data. In this paper, we propose an image deblurring method based on denoising diffusion models that can leverage the Lidar depth guidance and does not require training data with paired Lidar depth maps. We also present the first dataset with real blurred images with corresponding Lidar depth maps and sharp ground truth images, acquired with an Apple iPhone 15 Pro, for the purpose of studying Lidar-guided deblurring. Experimental results on this novel dataset show that Lidar guidance is effective and the proposed method outperforms state-of-the-art deblurring methods in terms of perceptual quality.