arXiv subjects

Isabelle Augenstein

Publications and source records attributed to Isabelle Augenstein.

At least 19 recordsLinked to original sources

arXiv2026

When Should I Trust This? Designing and Evaluating Factuality Indicators for LLM-based Chatbots

Generative AI systems increasingly mediate how people seek information, yet communicating AI factuality to laypeople remains challenging. Through a mixed-methods investigation involving interviews (N=8), a survey (N=1,000), and an experiment (N=5,151), we examine how people evaluate AI-generated information for gardening, mathematical, and medical queries, and how factuality indicators (confidence and sources) shape perceptions. Our interviews and survey show people selectively verify AI responses using external and AI-provided evidence, moderated by contextual factors. Although confidence-indicators affected trust and accuracy perceptions, people hold diverse, ambiguous interpretations of AI confidence. Sources were viewed as crucial, yet users raised concerns about source reliability and verification effort. Our experiment showed that confidence and source-based indicators influenced initial perceptions, but effects disappeared once participants learned the response's correctness. Medical queries elicited skepticism regardless of indicator. Findings motivate helping users interpret model confidence and integrating source information into the interface to support calibrated trust in AI-supported information-seeking.

cs.HC↗

arXiv2026

A Reality Check of Inference-Time Context Augmentation Methods

Large language models (LLMs) often fail to utilise the context given in the input, resulting in hallucinations or a fallback to parametric knowledge. Context augmentation methods aim to improve model performance by modifying the input, from inserting filler tokens that add computational time to adding relevant content such as elaboration of a context during inference. However, these methods have been evaluated in isolation and only by their performance gains, leaving it unclear how they compare and whether they reliably improve performance. In this paper, we present an evaluation framework spanning five methods that differ in how informative the inserted content is and apply them at the sentence level across seven models spanning multiple sizes on three datasets. We find that accuracy gains are not always genuine improvements but undirected perturbations: the methods correct some predictions while breaking others, particularly when the inserted text allows only for more computational time. Our analysis reveals that methods that bring no additional content primarily change model predictions near the decision boundary. Our findings highlight the need to evaluate the reliability of the accuracy gains before adopting them in practice.

cs.CL↗

arXiv2026

Investigating Multi-Cultural Value Selection Biases in LLMs

Large language models (LLMs) serve users whose cultural identities are multi-layered, yet cultural evaluations commonly reduce identity to one label. In this work, we study LLMs under simultaneous identity cues, with a country persona in the system prompt and a second identity signaled in the user turn, either by naming a country (explicit cue) or through the query language (implicit cue). We test World Values Survey (WVS) multiple-choice probing and open-ended sociopolitical generation. In multiple-choice probing, language cues shift answers about three times more than explicit cues but away from the associated country's distribution, and explicit cues do not consistently move answers toward the named country. Across 1.6 million open-ended responses, naming a country closes 30-44% of the gap to the model's own responses for that country, and explicit cues produce more differentiated value changes than language cues. Our results suggest that country and language cues should be evaluated as distinct signals rather than interchangeable proxies for culture.

cs.CL↗

arXiv2026

Output Vector Editing for Memorization Mitigation in Large Language Models

Large language models can reproduce training text, creating privacy, copyright, and security risks. We propose Output Vector Editing (OVE), a post-hoc method for suppressing text continuations that a model reproduces from a given prefix. Under superposition, a neuron can contribute to multiple features. Yet some neuron-level mitigation methods suppress localized neurons by zeroing their activations, removing each neuron's entire contribution. OVE instead edits their output vectors, combining (i) two-stage localization of MLP layers and neurons with (ii) full-vocabulary least-squares updates. Its multi-anchor variant, OVE-MA, adaptively selects continuation positions expected to provide stronger suppression relative to the required weight change. We evaluate recovery and locality across four models (360M--7B parameters), using autoregressively validated continuations and recovery tests with greedy decoding, beam search, and sampling. On 634 OLMo-7B pairs, OVE-MA with at most five output-vector updates per request leaves 24.45\% recoverable in these tests, compared with 32.02\% for the closest prior post-hoc editing method, REVS, while using approximately 75 times fewer update operations. These results show that adaptive position selection can concentrate effective suppression into a small number of weight updates. OVE edits weights without gradients and requires no additional intervention during inference.

cs.CL↗

arXiv2026

Unintended Side Effects of Value Training in LLMs

As LLMs are increasingly deployed in value-sensitive contexts, aligning them with human values has emerged as a post-training objective in NLP, aimed at instilling additional values and guiding model behaviour. However, current evaluations remain narrow. Because values are interconnected, it is unclear how fine-tuning for specific values alters unrelated capabilities and behaviours. In this work, we adapt four instruction-tuned LLMs across individual Hofstede's value dimensions and evaluate them on both cross-value transfer and broader behaviours. We find that value-specific training produces unintended side effects on untrained values, risk preferences, social cognition, and safety traits. Consequently, we suggest that value alignment after post-training should incorporate broader behavioral evaluations, and we highlight the need to investigate where additional value training might best fit into the pipeline.

cs.CL↗

arXiv2026

Shaping the Document, Not the Model: Optimized Suffixes for Context Following

When a user supplies a document and wants the answer drawn from it, a language model is expected to follow it even when it contradicts its parametric knowledge. Prior work has focused on making the model rely on context by steering activations, contrasting logits during decoding, or instructing it in plain language. We ask whether the document alone, with no instruction written into it and no access to the model at inference time, is enough. We optimize a document suffix by discrete token search, starting from the answer to a single question and extending the objective to a set of passages, to multi-hop questions that follow from the planted fact, and to later turns of a conversation. On counterfactual passages that contradict parametric knowledge of the model, the suffix flips nearly all items the model resists across three instruction-tuned model families, in most settings outperforms the decoding baseline and the strongest instruction we could write, and matches activation steering. Controls show that the suffix largely acts through the passage rather than injecting the answer, and a single suffix per model transfers to unseen passages. Each extended objective carries the effect to its target, so the objective, not only the document, sets how far the effect reaches.

cs.CL↗

arXiv2026

Mitigating Cross-Lingual Cultural Inconsistencies in LLMs via Consensus-Driven Preference Optimisation

Multilingual large language models (MLLMs) often answer culturally ambiguous questions inconsistently across languages, even for a fixed user persona -- e.g., Shakespeare in English but Cervantes in Spanish for the same British user. To robustly quantify this Cross-lingual Cultural Inconsistency, we introduce Singleton Fleiss's $κ_S$, a metric mathematically resilient to hallucinations. For mitigation, we propose Cross-lingual Cultural Consistent Preference Optimisation (C-3PO), a consensus-driven alignment framework. C-3PO raises $κ_S$ by 0.07--0.10 over unaligned models, above every prompting and representation steering baseline averaged across personas, whilst largely preserving explicit user identities, cultural neutrality and intrinsic cultural knowledge; where it does not, a soft-consensus variant recovers most of the cost. Although trained only on multiple-choice questions, C-3PO also improves consistency in open-ended generation. Empirical evaluations demonstrate this inconsistency is more pronounced for lower-resource languages like Indonesian and Persian. Finally, early decoding of intermediate layers reveals that MLLMs implicitly personalise outputs towards the prompt language's stereotypical culture as forward-pass representations stabilise.

cs.CL↗

arXiv2026

Not What, But How: A Framework for Auditing LLM Responses across Positioning, Generalization, Anthropomorphism, and Maxims

Large language models (LLMs) are being increasingly used to answer subjective, information-seeking questions, where users are sensitive to how responses are communicated, not just whether the answers are correct. Existing LLM evaluations for subjective cultural queries largely focus on factual correctness, ignoring how the response is framed. To this end, we introduce FRANZ, an automated FRAmework for respoNse characteriZation to conduct communicative audit of LLM responses along four dimensions: cultural positioning, use of generalizing language, anthropomorphic cues, and adherence to conversational maxims. To enable this evaluation, we contribute SQUARE - a corpus of 376k subjective questions sourced from 57 subreddits, and mapped to 7 countries and 19 question categories. We demonstrate FRANZ's applicability by scoring responses from three open-weight LLMs. We observe that LLMs show statistically significant differences in the frequency with which they employ each response characteristic. Unlike single-dimensional audits, FRANZ reveals that insider positioning and anthropomorphism are positively coupled, with the degree of coupling varying by country, providing a diagnostic lens for identifying framing divergences.

cs.CL↗

arXiv2026

The Reliance Hypothesis: Chain-of-Thought Faithfulness Rises as Language Resource Level Falls

Chain-of-Thought (CoT) faithfulness matters for monitoring, explaining and safely deploying large language models, yet it has been studied mostly in English. Languages differ in level, the amount of text available to learn from, so models know less in lower-resource languages, which may change how much they rely on their reasoning. To our knowledge, we are the first to relate CoT faithfulness to both resource level and task type (knowledge- or reasoning-intensive). On parallel multiple-choice questions from Multilingual Massive Multitask Language Understanding (MMMLU) in eight languages, faithfulness rises as resource level falls on knowledge-intensive tasks but not on reasoning-intensive ones, a gap that points the same way across four models and several faithfulness measures. We propose, and find support for, the Reliance Hypothesis: models rely more on their reasoning when they cannot recall the answer, as in lower-resource languages on knowledge-intensive tasks. Model confidence partly explains the effect of resource level, but not the knowledge-reasoning gap. Languages in which models are less confident rely more on reasoning, yet less confident answers within a language do not, a Simpson's paradox. In lower-resource languages, models rely on their reasoning for more answers, not more strongly for each. CoT monitoring may therefore be most informative in lower-resource languages, where it is least often validated.

cs.CL↗

arXiv2026

Explaining the Roles of Experts in Mixture-of-Experts LLMs for Personalization

Personalization of Large Language Models (LLMs) is increasingly important for user-centric applications, yet prior methods mainly adapt dense LLMs, whose polysemantic neurons, each encoding multiple unrelated concepts, hinder localizing the components behind personalization, and thus understanding and controlling it. We propose to personalize Mixture-of-Experts (MoE) LLMs, which sparsely route each token to a few specialized experts, reducing polysemanticity and enabling such localization. We adapt three open-weight MoE LLMs on a large-scale personalization dataset and train crosscoders on paired activations of the personalized and non-personalized models to learn interpretable personalized latents, from which we identify the experts encoding distinct personalization features. The personalized MoE LLMs perform competitively with dense LLMs while activating only a fraction of their parameters, exhibit lower polysemanticity than dense LLMs with similar active parameters, and, within the same model family, yield personalized latents that are less shared with the non-personalized model. Causal interventions show that these latents suffice to recover the personalized model's behavior and that boosting their associated experts moves the non-personalized model toward it, revealing distinct routing paths for different personalization features.

cs.CL↗

arXiv2026

Claim Framing as a Driver of LLM Sycophancy

Sycophancy in Large Language Models, the tendency to align responses with user preferences over ground truth, has largely been studied using questions without a clear ground truth and with a single, plainly stated user opinion. We present a systematic study of which framings of a user's claim trigger sycophancy in factual question answering, where every answer is objectively verifiable without extra context. Evaluating 14 models on 42K questions, we express correct and incorrect user opinions through various claim framings over 8 dimensions and 21 types, and measure a broad set of opinion-driven behaviors. We find that all models are susceptible to user opinions regardless of size. Models are most influenced by how confidently an opinion is stated, whom it is attributed to, and whether it is presented as shared between the user and the model. In contrast to prior work, framings based on the user's expertise, authority over the model, or social pressure matter little in our factual setting. Larger models resist incorrect opinions better but are no better at recovering from their own errors. Our results call for sycophancy evaluations that probe how an opinion is framed, not merely whether one is present, and that measure multiple behaviors beyond a single flip rate.

cs.CL↗

arXiv2026

From Concept Alignment to Causal Grounding: An Intervention Test of Chain-of-Thought Faithfulness

Chain-of-thought (CoT) can sound plausible yet be unfaithful to the model's underlying reasoning. Most prior work probes CoT faithfulness through input--output behavior or input attributions, leaving internal computation largely underexplored. We instead cast faithfulness as internal concept grounding: Does a large language model's (LLM) CoT reasoning engage the same internal concepts that support the LLM's direct prediction, and do the shared concepts causally drive its answer? Encoding a prediction pass and a CoT pass with a single shared sparse autoencoder (SAE), a reliable approximator of the latent concepts LLMs use, makes their internal concepts directly comparable. We introduce three correlational metrics of concept-level alignment and a causal metric, $Δp$, which ablates the shared concepts and measures the drop in answer probability. Across five LLMs and four datasets, concept alignment is generally high, as indicated by the correlational metrics; yet these only identify which concepts are shared, not how much they causally contribute. $Δp$ fills this gap: causal faithfulness varies substantially with model depth, peaking at mid-to-late layers rather than the final ones, and model scale reshapes the layer-wise profile. Moreover, causally important shared concepts are not always verbalized in the CoT. These dissociations suggest that faithfulness cannot be reliably assessed from surface-level or representational correspondence alone; assessing it requires causal tests of whether the internal concepts underlying a CoT actually drive the model's prediction.

cs.CL↗

arXiv2026

BiasGym: A Simple and Generalizable Framework for Analyzing and Removing Biases through Injection

Understanding biases and stereotypes encoded in the weights of Large Language Models (LLMs) is crucial for developing effective mitigation strategies. However, biased behavior is often subtle and non-trivial to isolate, even when deliberately elicited, making systematic analysis and debiasing particularly challenging. To address this, we introduce a simple, cost-effective, and generalizable framework \texttt{BiasGym} for reliably injecting, analyzing, and mitigating conceptual associations of biases within LLMs. \texttt{BiasGym} consists of two modules: \texttt{Inject}, which injects specific biases into the model via token-based fine-tuning while keeping the model frozen, followed by two debiasing methods that leverage these injected signals to identify and reliably suppress (\texttt{Scope}) or \texttt{Steer} the components responsible for biased behavior. Our framework enables consistent bias elicitation for better localization of bias conceptual association in the model space, supports targeted debiasing without degrading performance on downstream tasks, and generalizes to biases unseen during fine-tuning. We demonstrate the effectiveness of our proposed framework in reducing real-world stereotypes (e.g., people from Italy being `reckless drivers'), showing its utility for both safety interventions and interpretability research.

cs.CL↗

arXiv2026

Multi-Step Knowledge Interaction Analysis via Rank-2 Subspace Disentanglement

Natural Language Explanations (NLEs) describe how Large Language Models (LLMs) make decisions by drawing on external Context Knowledge (CK) and Parametric Knowledge (PK). Understanding the interaction between these sources is key to assessing NLE grounding, yet these dynamics remain underexplored. Prior work has largely focused on i) single-step generation and ii) modeled PK--CK interaction as a binary choice within a rank-1 subspace. This approach overlooks richer interactions and how they unfold over longer generations, such as complementary or supportive knowledge. We propose a novel rank-2 projection subspace that disentangles PK and CK contributions more accurately and use it for the first multi-step analysis of knowledge interactions across longer NLE sequences. Experiments across four QA datasets and three open-weight LLMs demonstrate that rank-1 subspaces struggle to represent diverse interactions, whereas our rank-2 formulation captures them effectively, highlighting PK alignment for supportive interactions and CK alignment for conflicting ones. Our multi-step analysis reveals that hallucinated generations exhibit strong alignment with the PK direction, though context-faithful generations maintain a more balanced alignment between PK and CK.

cs.CL↗

arXiv2026

What and Whose Knowledge? Measuring Epistemic Diversity in Large Language Models

Large language models (LLMs) are increasingly used as primary knowledge sources, yet their epistemic diversity - defined as the diversity of real-world claims in their outputs - has never been measured. Low epistemic diversity would pose a risk of knowledge collapse as homogeneous LLMs mediate a shrinking in the range of accessible information over time. The dominant paradigm is that overall LLM diversity is low, but this is always with respect to a single point in time, with no reference baseline or consideration for variation across countries. We address this gap in knowledge by performing the first systematic study of epistemic diversity in LLMs across time and cultural context, testing 27 LLMs on 155 topics covering 12 countries, resulting in 1.7M responses and 70M individual claims. We find that epistemic diversity has increased substantially over the past three years, a positive counter to recent diversity pessimism. However, despite progress, we find that every system is less diverse than a search baseline. This gap is not uniform: RAG can improve diversity, while large models are counterintuitively less diverse than smaller ones. Moreover, LLM parametric knowledge systematically reflects English over local-language knowledge for country specific topics. Together, these results demonstrate that while progress on epistemic diversity is tangible, it is insufficient and unevenly distributed.

cs.CL↗

arXiv2026

Do We Still Need Humans in the Loop? Human vs. LLM Annotation in Active Learning for TikTok Hate Speech Detection

Annotating data remains a costly bottleneck for supervised NLP. Active learning (AL) reduces the number of human labels needed by selecting only the most informative instances, while instruction-tuned LLMs attack the same bottleneck from the other side, making labels cheap enough to annotate entire corpora. This raises two questions: can LLM labels replace human labels within the AL loop, and does AL remain necessary when entire corpora can be cheaply labeled? We investigate both by training supervised hate speech classifiers on a new dataset of 278K German political TikTok comments, comparing human and LLM annotation under matched conditions. LLM annotation at scale outperforms human-supervised classifiers at roughly one-tenth the cost, for both a closed-source (GPT-5.2) and an open-weight (Qwen3.5-122B-A10B) LLM, and the advantage is robust under soft-label evaluation. It hinges on the annotation interface: only a two-question decomposition mirroring the human annotation task unlocks it. AL provides no reliable advantage over random sampling in our prefiltered pool. Error structure depends on the LLM: only GPT-5.2 matches the human FP/FN balance, while other variants over-flag border-control and economic-competition discourse. Humans remain essential as evaluators; for training labels, the question shifts to which LLM, which interface, and what shape of pool.

cs.CL↗

arXiv2026

MMMMM: A Unified Taxonomy for Investigating the Mechanisms of Multilingual MultiModal Misinformation

Multimodal misinformation on social media is highly prevalent, potent, and harmful, yet difficult to detect and counter, and still poorly understood compared to its text-only counterpart. Research on the properties and deceptive strategies of multimodal misinformation is hindered by a lack of taxonomies grounded in real-world contexts and by the limitations of current multimodal machine learning models, which prevent the automation of annotation and analysis at scale. We address these shortcomings in three steps. First, we collect a large-scale, high-quality dataset of real-world misinformation instances from Twitter/X in seven languages. Second, we develop a novel, comprehensive taxonomy of multimodal misinformation grounded in an in-depth qualitative analysis of the data and prior theoretical work. Finally, we operationalise the taxonomy through an automated multi-step annotation pipeline using a Vision-Language Model (VLM), and perform human-validation. Our novel approach leads to previously undocumented insights about how social media users combine images with text to spread misinformation in the wild, e.g., that AI-generated content is particularly prevalent in technology and science, while vaccination misinformation disproportionately utilises images from news outlets to assert credibility. Our method and findings provide guidance for targeted approaches for detecting multimodal misinformation, and suggest that mitigation efforts should be developed and applied strategically rather than uniformly.

cs.CY↗

arXiv2026

CulTrace: Tracing Internal Cultural Reasoning in Large Language Models

The growing deployment of large language models (LLMs) across diverse cultural contexts necessitates a deeper understanding of models' hidden representations of different cultures. Prior work has evaluated cultural awareness in LLMs by analysing their outputs. This approach overlooks how cultures are represented within the model parameters, missing why models generate incorrect responses. To bridge this gap, we propose CulTrace, a mechanistic interpretability-based method that probes the internal representations of LLMs for cultural knowledge. With CulTrace, we inspect how cultural knowledge is processed across layers and how it is integrated during cultural QA. We find a consistent staged trajectory of cultural reasoning. Models first engage with the question's domain, then resolve the relevant culture, and finally narrow in on an answer. We also demonstrate that models' cultural reasoning is imbalanced, showing delayed relevant culture resolution and more confusion with less-represented cultures.

cs.CL↗