Explainable AI Methods

Time Series Anomaly Detection has received increasing attention, driven by the growing availability of complex time series data. This surge has led to the development of numerous detection methods, as well as a variety of benchmarks aimed at thoroughly evaluating their performance. However, most existing detectors remain largely agnostic to domain context, overlooking explainability and interpretability. One of the main reasons for this gap is that current benchmarks primarily focus on detection accuracy, and only few of them evaluate spatial explainability. Moreover, no benchmark currently provides sufficiently rich semantic annotations to support the generation of human-understandable interpretations of anomalies. To address these limitations, we introduce SHAD (Scality High-dimensional Anomaly Detection benchmark), a fully annotated benchmark composed of 215 multivariate, high-dimensional time series collected from real-world distributed cloud storage systems operated by Scality. The proposed dataset includes rich contextual information, covering three families of anomalies with varying degrees of severity. As further contribution, we provide a foundation for future work by evaluating baseline methods for Detection, Explainability, and Interpretability, covering all stages of a TSAD pipeline. For Detection, we benchmark a wide range of existing anomaly detectors, testing their effectiveness on the proposed real-world dataset. Then, we consider explainability by evaluating whether measuring the contribution of each dimension in the generated anomaly score can provide accurate anomaly attributions. Finally, for interpretability, we investigate the effectiveness of frozen LLM baselines in localizing and interpreting anomalies.

Existing synthetic image evaluators typically provide only a scalar quality score and do not identify the image regions that support it. We introduce VIEScore2, a unified evaluator for image generation and editing tasks with optional conditioning images. VIEScore2 represents an image as an N x N grid and jointly predicts quality scores and defect locations in a single model pass. Its text-native grid representation provides a common interface for heterogeneous spatial supervision and enables directly verifiable post-training objectives. We train on 38K examples spanning score-only, localization-only, and joint supervision across generation and editing tasks. Starting from supervised fine-tuning, we further apply GRPO to improve defect localization using rewards that combine cell-level Dice overlap, score accuracy, and output-format validity. A parameter-free parser converts the structured predictions into readable explanations. On the primary suite, VIEScore2 achieves an overall-score SRCC of 0.601, compared with 0.491 for Gemini-3-Flash, the strongest zero-shot general-purpose VLM baseline under matched inputs. For defect localization, VIEScore2 outperforms both general-purpose VLMs and specialized spatial evaluators on three of six benchmarks in per-image grid IoU and ranks among the top three on five, including datasets beyond its training sources.

When a graph neural network (GNN) explainer produces an unexpected attribution on a molecule, the attribution alone cannot reveal whether the explainer has failed or the model has learned a shortcut. We introduce WOMBAT, a benchmark of 14 whitebox GNNs, each with message-passing weights set by hand to detect a specific SMARTS motif. Each model's decision rule is known by construction, providing attribution ground truth against which explainer errors can be identified and studied. We validate the models on millions of PubChem molecules and evaluate post-hoc explainers including GNNExplainer, PGExplainer, and Integrated Gradients. Guided by our qualitative analysis, we construct a model that causes Integrated Gradients to spread attribution across the graph, even though the model reliably detects the intended motif. We release the dataset, models, and evaluation code to help researchers in the development of newer XAI tools for GNNs.

Machine-learning surrogates accelerate physical simulation, but lower prediction error need not coincide with lower error in physically relevant flow statistics. We examine this question for flow around a NACA4418 airfoil using paired computational-fluid-dynamics simulations and experimental particle-image-velocimetry measurements. A mean-preserving input intervention removes velocity fluctuations from selected regions of observed flow histories. Across four neural operators, removing fluctuations from the most energetic 10% of valid observed cells changes forecasts more than equal-area random removal. Because the masks are not matched for removed fluctuation energy, this contrast measures sensitivity, not independent evidence of physical importance. Separately, a CNO has lower velocity-field error but substantially higher two-component fluctuation-energy error than the reference on both analysis subsets. An output attenuation stress test also demonstrates disagreement between benchmark errors and domain-summed fluctuation energy. These single-benchmark results motivate reporting complementary physical diagnostics alongside aggregate prediction scores; they do not establish counterfactual physical correctness.

Synthetic speech attribution aims to identify the generative system responsible for a speech signal, but current approaches typically rely on black-box neural networks that provide limited insight into their decisions. This work investigates prototype-based networks as an interpretable alternative, where predictions are grounded in comparisons with representative training examples. We adapt ProtoPNet to spectrogram-based speech representations and evaluate the proposed framework on the MLAAD dataset under closed-set, cross-lingual, and open-set conditions. Experiments show that prototype-based reasoning achieves competitive or improved attribution performance compared with the baseline while enabling example-based explanations. These results highlight that interpretability and performance can be jointly achieved in synthetic speech attribution through prototype-based modeling.

Natural language autoencoders translate a language model's internal activations into readable explanations. Explaining every token position is costly. Which positions should an auditor inspect to understand a potential threat? We study this question across $4.7$ million explanations on prompt injection and concealment. We compare signals from model computation with a ranker trained only on chat structure. Chat structure usually selects more relevant explanations than the computational signals, without requiring a model forward pass for position selection. On three of four datasets, explaining just $5\%$ of positions retains nearly all of the success rate from explaining every position, where success means obtaining an explanation about the threat. The benefit varies with the audit task. We also show that pretrained verbalizers recover words that models have learned to conceal through fine-tuning, without additional verbalizer training. These results identify where auditors can concentrate explanation generation and show that useful explanations can extend beyond the model a verbalizer was trained to describe.

Can interaction formats and textual scaffolds help large language model (LLM) agents make better decisions, and do better decisions come with better explanations? We study these questions in auctions and matching, multi-agent environments with explicit rules and known optimal strategies. These settings let us vary how a decision problem is presented while retaining a benchmark for evaluating behavior. Drawing on human-motivated theories of simplicity, we compare interfaces that elicit a complete bid or ranking with sequential interfaces that make safe choices easier to identify. We then hold the interaction format fixed and vary reasoning scaffolds and rule descriptions. Across four model families, the ascending auction interface substantially reduces bid deviations. The matching comparison also shows why sequential responses require different error accounting from complete rankings. Laying out payoff contingencies and explaining why truth-telling is safe also improve choices, whereas prompts to plan through matching rounds or form beliefs about opponents worsen play overall. In auctions, these behavioral gains are not accompanied by corresponding improvements in measured verbal indicators of strategic understanding in the agents' short stated plans. Other prompts change those indicators without improving bids. Our findings suggest that human-motivated theories of simplicity can inform the design of decision environments for artificial agents. They also show why scaffolds should be evaluated through realized choices as well as explanations: improvements in one need not appear in the other.

Scientific figures are designed to communicate information visually, yet MLLMs typically explain them by translating their visual content back into text. This requires readers to manually map the resulting explanations back to the figure. Inspired by how people present visual information, we introduce FigAct, a framework that transforms static scientific figures into question-conditioned visual presentations by acting directly on their existing graphical elements. Like a human presenter, FigAct generates a sequence of short narrations, grounds each narration in the corresponding visual evidence, and applies visual actions to guide the viewer's attention. We develop a hierarchical search strategy for efficient element localization, reducing token usage by approximately 40$\times$. We further train FigAct-8B using three task-specific rewards for grounding accuracy, search efficiency, and rendering quality. We further build a human-verified benchmark from figures in real-world scientific papers to evaluate the ability of MLLMs to generate grounded visual explanations. Our results demonstrate the effectiveness of FigAct and show that treating scientific figures as presentation canvases makes explanations clearer and easier to follow.

Automated detection of LLM-generated texts (LGTs) is critical, yet dedicated detectors often struggle to generalize across domains and models. While general-purpose LLMs offer flexible zero-shot authorship classification with explanatory rationale, their detection behavior, especially regarding self-detection versus cross-detection across model generations, remains poorly understood. We systematically evaluate 15 LLMs spanning three model generations as both generators and detectors. Using a benchmark of 1,000 human-written texts and 15,000 LGTs (1,000 per model), we collected over 233,000 binary classifications alongside natural-language explanations. Our results reveal that detection efficacy is primarily driven by detector capability rather than generator provenance, although outputs from newer generators remain notably harder to detect. Crucially, statistical comparisons show no systematic advantage or disadvantage for self-detection across models. Error analysis further exposes generational bias shifts: first-generation detectors under-detect LGTs (high false-negative rates), second-generation detectors over-flag human texts (high false-positive rates), and the latest models achieve balanced trade-offs. Finally, we highlight significant inconsistencies in how different LLMs apply textual cues to justify their decisions. Code: https://github.com/hyyuan/detect-llm-generated-texts.

Reward models (RMs) are a key component of large language model post-training, providing reward signals for subsequent reinforcement learning. However, conventional discriminative RMs typically output only scalar scores, making it difficult to identify the response behaviors associated with their scoring decisions. Existing interpretation methods often rely on predefined high-level attributes and require repeated counterfactual interventions for each response pair to validate candidate explanations, lacking a closed-loop mechanism that uses RMs' feedback to train a reusable explainer. To address this, we propose RewardExplainer, a framework that obtains feedback from the target reward model through counterfactual rewriting and uses this feedback to further optimize the explainer. RewardExplainer generates open-ended, atomic, and intervenable natural-language scoring mechanisms, making explanations more concrete, readable, and actionable. It further converts counterfactual feedback into preference supervision, enabling the explainer to more faithfully capture the target RM's scoring preferences and sensitive behaviors than single-pass generation. Extensive experiments across multiple target RMs and explainer backbones show consistent improvements. Beyond interpretation, we use the generated mechanisms to identify potential bias patterns and construct targeted debiasing data for fine-tuning the reward model, improving robustness on reward-hacking benchmarks.

Natural Language Autoencoders (NLAs) produce unsupervised text explanations of a model's activations: a verbalizer describes an activation and a reconstructor learns to recover it from this text. Under the established point-reconstruction NLA training recipe, explanations become more useful for predicting model behavior while also increasingly introducing unsupported details and exhibiting writing defects. To assess these changes separately, we introduce a standardized evaluation framework for unstructured NLA explanations, measuring information recoverable from explanations, contextual support for their claims, and writing quality. To address confabulation and writing defects, we move beyond predicting a single activation: explanations can distinguish distributions of possible activations even when their means and optimal point-reconstruction rewards are identical. We introduce Flow-NLA, which models the distribution of activations compatible with an explanation and trains the verbalizer using a diffusion likelihood bound. Across Qwen, Gemma, and Apertus, this richer signal retains the utility gains of point reconstruction while curbing the growth of confabulation and writing defects, opening up a direction for improving activation-derived training to encourage more informative, supported, and readable explanations. Code and evaluation prompts will be made publicly available upon acceptance.

Time-series explainability remains challenging because discriminative information is often encoded in latent frequency or time-frequency features rather than in the raw signal itself. Existing attribution methods typically operate either in the time domain or in a fixed transform domain, limiting their ability to capture salient information across different representations. We propose XACT, a general framework that learns sparse attribution masks over coefficients from arbitrary invertible time-frequency transforms. We evaluate the framework on the STFT, the continuous wavelet transform, and the discrete wavelet transform. In addition, we extend the virtual inspection layer approach from the STFT to both wavelet transforms, enabling LRP to generate explanations in these representations. On a synthetic dataset, XACT produces precise explanations and is less prone to highlighting spurious features than the tested baselines. Across two real-world datasets, XACT produces sparse and structured explanations, although no method performs best across all quantitative evaluation criteria. These results demonstrate that learning explanations directly in time-frequency representations offers a flexible approach to interpreting deep-learning models for time series data.

Post hoc explanation methods such as SHAP and LIME are widely used to interpret text classifiers, but their visualizations are mainly designed for left-to-right languages. When applied to right-to-left (RTL) languages such as Urdu, Arabic, Persian, and Hebrew, the attribution values remain mathematically valid, while their visual presentation fails. Tokens appear out of sequence, connected letterforms break apart, and plot layouts do not follow the natural reading direction. This study addresses this gap as a visualization problem rather than a limitation of the explanation methods themselves. We present SHAP-RTL, a rendering layer that corrects reading direction and script shaping in SHAP and LIME visualizations, with per-language font selection, while preserving the original attribution values, feature ordering, and model outputs. The approach is evaluated on Urdu, Arabic, Hebrew, and Persian hate and offensive-language datasets using TF-IDF and logistic regression classifiers. Rendering correctness is measured by an OCR round trip over 200 feature words per language. Default rendering yields character error rates of 0.820 to 0.979, meaning the label no longer carries its token; the common reshape-and-reorder workaround fails for Urdu at 0.998, worse than no correction; and the Matplotlib 3.11.0 text rewrite inverts that workaround, while SHAP-RTL remains correct under both versions. The framework also verbalizes the same attributions as short contextual explanations in the reader's language, constrained to the identified features. Evaluation in this paper concerns rendering correctness; assessment of the generated explanations is left to future work. The study highlights the importance of language-aware visualization in making post hoc explainability more accessible across different writing systems.

Visual attribution methods typically explain predictions using pixels, superpixels, or regular patches. These representations can localize important regions, but provide limited information about their structure. We introduce MorphoSHAP, a model-agnostic post-hoc method that instead uses morphological shapes as the players of a Shapley attribution game. Using the Tree of Shapes, each shape is described by its scale, geometry, and signed contribution, providing explanations of where the evidence lies, what type of structure carries it, and how strongly it affects the prediction. This shared morphological vocabulary enables spatial, textual, and global class-level explanations beyond image-specific heatmaps. To the best of our knowledge, MorphoSHAP is the first SHAP-based image attribution framework to combine these different forms of explanation. Across five diverse datasets and three architectures, MorphoSHAP achieves strong insertion/deletion performance and outperforms competing attribution methods on several benchmarks. Finally, a user study shows that MorphoSHAP provides explanations that are easy to use and are preferred over standard attribution baselines.

We present a collaborative streaming anomaly detection system for high-speed data streams that explicitly integrates human analysts into the decision loop. The system combines heterogeneous detectors and aggregates their outputs through a normalization-based weighted consensus, complemented by artifact-aware rules to stabilize anomaly scoring under deployment. To improve interpretability, it derives surrogate models that approximate the ensemble consensus and expose human-readable sensor conditions associated with anomalous behavior. Analysts can actively intervene by reviewing anomaly episodes, adjusting consensus behavior, and refining surrogate rules used for anomaly prediction, producing a human-adjusted ensemble. We evaluate the approach on an industrial stream with 260,000 events and 3 anomalous episodes, showing robust detection and actionable human-AI interaction.

Deploying efficient neural networks is essential in resource-constrained environments, yet compact models often sacrifice interpretability - a critical in safety-critical domains such as autonomous driving and medicine. This study investigates whether Knowledge Distillation transfers the spatial feature attribution of a large teacher network to a compact student. To assess the influence of the KD scheme on interpretability, we distill a ResNet-152 teacher into a ResNet-34 student on ImageNet-1K across five configurations by systematically varying the distillation temperature and soft-label loss weight. Models are evaluated on top-1 accuracy, along with two interpretability metrics: Relevance Mass Accuracy and Relevance Rank Accuracy. These metrics are computed via Grad-CAM heatmaps benchmarked against ground-truth object masks. Our results show that top-1 accuracy ranges from 71.6% to 74.0%. For Grad-CAM, RMA ranges from 7.7% to 9.7% and RRA from 7.3% to 10.1%; for Guided Grad-CAM, RMA ranges from 16.1% to 18.6% and RRA from 15.9% to 21.5%. Interpretability proves far more sensitive to the soft-label weight than to the temperature: keeping the student anchored to hard labels preserves both accuracy and coarse localization, whereas weighting the teacher heavily degrades both. Fine-grained attribution, however, fell below the undistilled baseline in every configuration tested, indicating that logit distillation transmits where a model attends more readily than the pixel-level structure of that attention. We evaluate 12 cross-architecture combinations of convolutional and transformer-based models, revealing that the inheritance of fine-grained spatial reasoning is fundamentally bottlenecked by the student's intrinsic structural biases. To our knowledge, this is the first application of this interpretability-aware evaluation framework - previously used for neural network pruning - to KD.

Automated pain recognition from facial expression could make continuous welfare assessment practical in sheep, but adoption depends on trust: a stockperson cannot act on a score that arrives without justification. We ground a model in the Sheep Pain Facial Expression Scale (SPFES) by letting each detected facial region attend over text embeddings of the clinical descriptors and then test whether the resulting explanations mean anything. They do not. Ablating an entire descriptor changes the predicted logit by about $10^{-4}$, and the most-attended cue agrees with the predicted pain level in only $32.6\%$ of regions, although the attention maps, the learned gate, and the generated text all proposed otherwise. We therefore remove the appearance bypass with a concept bottleneck whose classifier reads only SPFES concept scores, supervised by per-region state annotations that image-level pipelines discard. This costs $0.05$--$0.10$ in Cohen's $κ$ but yields concepts that are demonstrably learned: minority pain-indicating states are recovered at $3.5$--$8.3\times$ their base rates, and the ear and eye severity orderings emerge without severity supervision. Removing the supervision alone leaves $κ$ unchanged while concept accuracy falls to $0.109$, showing that architectural necessity does not imply semantic validity. We also show that pooled concept accuracy is misleading under clinical imbalance and provide a cross-validated, protocol-matched benchmark of seven methods on this dataset.

Explanations of machine learning models are usually judged by criteria that are hard to compare. We propose a simpler test: if an explanation really describes how a model uses its features, it should be possible to rebuild the model's predictions from it. We turn each explanation into a predictor by reading each feature's effect and adding them up, and measure how well that predictor reproduces the model on unseen data. Nothing is fitted, so the score reflects the explanation itself. The test applies to any explanation that can be written as a function of the features; we demonstrate it on partial dependence plots (PDP), accumulated local effects (ALE), SHAP and LIME. We prove that summing partial dependence curves gives the best possible additive summary of a model when its features are independent, and that this fails when they are dependent. Across 13 real datasets and 9 synthetic designs and four model families, which method scores best depends entirely on feature dependence: where features are independent SHAP is slightly worse than PDP, exactly as the theory predicts; on dependent real data SHAP leads. Some widely used quality metrics even prefer a damaged explanation to an intact one.

Counterfactual (CF) explanations identify changes that alter an input's classification. While existing methods produce realistic and low-cost CFs, they often fail to ensure feasibility, by suggesting non-constructive modifications or incompatible with future changes (e.g., changing an individual's race to secure a job offer). We introduce a refinement of CF explanations that explicitly enforces feasibility. Our approach is the first to efficiently generate CFs that are realistic, low-cost and feasible. We accommodate both hard feasible constraints, specified by domain knowledge users, and soft feasible constraints, inferred automatically via causal inference from the dataset. Our method, Feasible Counterfactual Explanations (FCx), is based on a modified Variational Autoencoder (VAE) optimized with a multi-factor loss function. We measure the cost of a change based on the absolute change in values (proximity) as well as the number of features changed (sparsity) while realism is measured based on the LOF for density estimation, guaranteeing that CFs reside in densely populated regions. Extensive experiments on four public datasets show that our approach matches state-of-the-art performance across multiple metrics while guaranteeing feasibility.

Formal explainability provides mathematically grounded justifications for individual predictions. However, abductive explanations often exceed human cognitive limits by involving too many features, while probabilistic relaxations have remained largely limited to categorical classification. We present a unified framework for probabilistic explainability based on sparse, anchored linear models, applicable to both binary classification and continuous regression. By mapping instances to the Boolean hypercube, our linear explanations strictly generalize subset-based approaches: they capture both the magnitude and direction of feature contributions while enforcing a prescribed sparsity budget $k$. We show that minimizing the relevance error for such explanations is \ClassNPPP-hard when the underlying model is a neural network, and we relate this intractable objective to a tractable surrogate---the fidelity error. For a parameterized family of local distributions, the relevance error of any $k$-sparse explanation is bounded by its fidelity error up to a multiplicative factor that remains small locally. We address the resulting empirical problem using two complementary approaches: a Mixed Integer Programming (MIP) formulation that yields provably optimal empirical solutions while maintaining polynomial sample complexity, and a polynomial-time Iterative Hard Thresholding (IHT) algorithm with provable approximation guarantees. Empirical evaluations show that, unlike state-of-the-art baselines such as LIME and MAPLE, our explanations satisfy both the anchoring and sparsity constraints by construction, while consistently achieving lower relevance error.

Large language models (LLMs) are increasingly deployed in applications involving interaction between agents, where their output plays a role in collective reasoning and decision-making processes. Despite significant research into the functioning of LLMs in such multi-agent systems, the processes of bias propagation in such systems are still a challenge. This work studies how biased opinions are propagated in the form of textual interaction in an environment of LLMs, in which a minority of agents maintain persistent extreme opinions, while the remaining agents iteratively update their beliefs through structured textual interactions. The findings show that even the presence of a small percentage of biased agents in such a system leads to significant shifts in the opinions of non-biased agents. It suggests that for the same percentage of biased agents, the shifts occur more quickly for the Llama~3.2 model when compared to a classical Friedkin-Johnsen (FJ) model. Further semantic analysis demonstrates that rhetorical consistency in textual explanations increases systematically with biased exposure and, importantly, is partially decoupled from numerical convergenumericalutral agents adopt the vocabulary employed by the biased agents even in configurations where their numerical opinion shifts remain moderate. The research helps explain how bias and language develop together in multi-agent language model ecosystems.

Vision-Language Models (VLMs) can produce Natural Language Explanations (NLEs) that sound plausible yet remain inconsistent with the visual evidence they cite. We present Explanation-Driven Counterfactual Testing (EDCT), an intervention-based protocol that extracts visual concepts cited in a model's explanation, applies verified minimal edits to them, and tests whether the resulting answer and explanation remain consistent with the edited image. Using this protocol, we create EDCT-Bench, a comprehensive benchmark spanning three complementary domains: knowledge-intensive visual question answering (OK-VQA), safety-critical driving (DriveLM), and 3D spatial reasoning (3DSRBench). Across the evaluated VLMs, EDCT reveals substantial faithfulness gaps, with models frequently producing responses inconsistent with verified visual changes. Finally, our fine-tuning study suggests that EDCT-generated counterfactuals provide high-impact training signals.

Large language models can easily generate explanations for their own outputs, but such self-explanations are not necessarily faithful to the model's behavior. We study this issue through counterfactual self-explanations, where a model minimally edits an input so that its own prediction changes. Across sentiment analysis and natural language inference, we evaluate ten instruction-tuned models from the LLaMA-3 and Qwen-2.5 families, measuring faithfulness, minimality, and alignment with human-annotated rationales. Our results show that model scale is the strongest determinant of explanation quality: larger models are substantially more likely to generate counterfactuals that flip their own predictions and target decision-relevant evidence. In contrast, the rationale-guided condition produces edit-minimal counterfactuals that are also more human-aligned. However, it does not consistently improve faithfulness. Overall, counterfactual self-explanations can provide useful behavioral evidence about model decisions, but their reliability depends strongly on model capacity and should be empirically validated rather than assumed.

Algorithmic recourse methods typically assume that a predictive model has access to all features of an individual. In practice, decisions are often made with partial information, because features are costly to acquire. Active feature acquisition addresses cost-constrained prediction, but existing methods are explanation-agnostic: prior work provides explanations only after acquiring additional features, rather than using explanations to drive acquisition. This work flips that and treats algorithmic recourse and feature acquisition jointly. We use Markov Blanket theory to unify counterfactual, semifactual, and alterfactual explanations and to characterize how available recourse grows as features are acquired. Building on this framework, we propose an Explanation-Driven Feature Acquisition (EDFA) method that selects features by explanatory value per unit cost. The framework is further extended with distribution-free validity guarantees for recourse issued from partial information, which signal trustworthy, lower-cost recourse, along with a lower bound on the calibration data required to certify them. Experiments on 7 publicly available datasets with neural network-based predictive models show that EDFA acquires substantially fewer features than state-of-the-art AFA baselines while maintaining comparable accuracy and yielding more decision-relevant, actionable recourse. The implementation is available on GitHub.

Short-video platforms have become a primary news source for the public, which has also enabled the widespread dissemination of fake news videos. We study the task of fake news video detection and explanation (FNVDE). Existing methods face two critical limitations. First, commonly used frame selection strategies may omit veracity-relevant cues or provide insufficient temporal context for understanding news videos. Second, prior methods neglect either multimodal understanding or evidence retrieval. To address these limitations, we propose NVKE-CEI, a unified system that integrates a news video keyframes extraction method (NVKE) and an FNVDE framework leveraging both content and evidence information (CEI). NVKE selects keyframes based on chronological changes in combined visual and OCR-text similarity. CEI employs two specialized LLM-based fact checkers (content-based and evidence-based) whose outputs are fused by a lightweight judge model. Extensive experiments show that NVKE-CEI outperforms state-of-the-art baselines while generating high-quality content-grounded explanations.

Evaluating the quality of explanations produced by explainable AI (XAI) methods remains challenging because existing approaches often rely on subjective human judgment, limiting reproducibility, scalability, and comparability between studies. We examine whether LLMs can serve as a reproducible and scalable mechanism to make comparative assessments of the quality of XAI explanations. We introduce XAI-Arena, an LLM-as-a-judge framework for scalable, reproducible, multidimensional, and stakeholder-sensitive evaluation of XAI explanation quality. XAI-Arena then allows us to compare XAI explanations along various dimensions, namely, perceived simplicity, clarity, task adequacy, trust calibration, actionability, transparency, faithfulness, and overall interpretability. We then benchmark XAI explanation methods across various datasets, machine learning models, and stakeholder personas. Human validation shows a strong positive association between LLM-generated and human ratings (Spearman's rho=.693, p<.001). Together, LLM-based evaluations can capture systematic differences in XAI explanation quality and provide a scalable and reproducible framework for comparative assessment of XAI explanations.

Brain-age prediction models are commonly evaluated by predictive accuracy, yet accurate predictions alone do not establish that a model relies on reproducible or neurobiologically supported mechanisms. Post-hoc explanation methods can expose these mechanisms, but existing workflows typically stop at diagnosis or require correction targets to be specified before model analysis. We propose XAI-Refine, an automated explanation-knowledge loop for brain-age prediction from resting-state functional connectivity. At each iteration, XAI-Refine consolidates complementary post-hoc analyses across repeated training runs into reliable, structured model explanations. It converts each reliable explanation into a neutral neurobiological question, retrieves and verifies relevant literature, and compiles the verified evidence into an admissible set in the same typed explanation space. The target for refinement is defined as the minimal projection of the current model explanation onto the admissible set induced by applicable verified knowledge. This revised explanation is then translated into a differentiable constraint while preserving the originating model variable, measurement operator, and applicable scope. Candidate updates are promoted only when multi-seed validation confirms target-directed explanatory movement, predictive performance remains within a prespecified guardrail, and non-target explanatory drift remains bounded. Experiments on functional-connectivity-based brain-age prediction evaluate predictive performance, explanation reliability, literature alignment, and target-specific model revision, illustrating a structured route from post-hoc analysis to evidence-guided model refinement.

Divergence-based regularization and Sharpness-Aware Minimization (SAM) are two prominent approaches for improving generalization in deep learning, both motivated by robustness to perturbations. However, their relationship has remained largely unexplored. Building on classical second-order expansions of $f$-divergences, we show that the two methods are locally consistent under parameter-space perturbations: both induce curvature-sensitive penalties, with divergence regularization yielding a Fisher-weighted quadratic form and SAM penalizing sharpness through the dominant Hessian eigenvalue. For negative log-likelihood objectives with exponential-family output distributions, this correspondence becomes especially transparent, since the Fisher and Gauss-Newton matrices coincide. We further show that the same local geometric perspective extends to input-space perturbations, where divergence-based regularization is defined through transformations of the input. In this setting, the regularizer induces a pullback quadratic form on the input space, providing a more general perturbation framework than standard SAM while preserving the same local sensitivity interpretation. To validate the analysis empirically, we use the asymmetric $α$-skew Jensen-Shannon divergence (JSD) family as a controlled testbed. Its local curvature coefficient scales as $α(1-α)$ and is maximized at the symmetric point $α=\tfrac12$, which recovers the standard JSD. Loss-landscape visualizations in the input-perturbation regime show that stronger induced curvature penalization is associated with flatter local minima. Experiments on four benchmark datasets further demonstrate that both accuracy and negative log-likelihood are consistently best near this regime of maximal curvature penalization.

Post-hoc explanation methods are widely used to inspect image classifiers, but their reliability depends on design choices that are often treated as implementation details. We study this issue for LIME on document image classification, focusing on the segmentation step that defines the interpretable units being perturbed. Standard image-based LIME typically relies on natural-image superpixels, which are poorly aligned with document structure such as text regions, layout blocks, and identification codes. Using RVL-CDIP, we compare Quickshift and SLIC with document-aware segmentations based on OCR bounding boxes and regular grids. Our results show that segmentation strongly affects explanation consistency, correctness, and local fidelity. Document-aware segmentations produce more stable and faithful explanations, require fewer perturbations to converge, and expose shortcut behaviour based on document identification codes, a known RVL-CDIP bias that superpixel-based LIME often obscures. These findings show that reliable post-hoc explanation requires domain-aware interpretable representations, and that segmentation should be treated as part of the explanation method rather than as neutral preprocessing.

LLM decision components that can operate within agent workflows often produce action-relevant recommendations or judgements together with explanations. Operators may use the named factors to monitor a system, diagnose errors, or decide when to escalate an output. Such use assumes that the explanations agree with the component's observable decision behaviour. We test two interpretations of the named factors: necessity, meaning that changing a factor would change the output, and sufficiency, meaning that retaining it while removing other changeable information would preserve the output. We evaluate these interpretations in two synthetic use cases: recommending advisors to clients and judging prompts for harmfulness or risk. Models return an output and the top three factors that most influenced it. Controlled black-box interventions estimate a necessity score for each factor by measuring how often changing it changes the output, and a sufficiency score by measuring how often retaining it preserves the output. Across eight models from the Claude, GPT, and Gemini families, the mean Spearman correlations between the cited ranking and the necessity and sufficiency scores are 0.349 and 0.354 for advisor recommendation, and 0.431 and 0.580 for prompt monitoring. Furthermore, an uncited factor scores above the lowest-scoring cited factor in 57.6% of advisor responses under necessity and 58.1% under sufficiency; the corresponding prompt-monitoring rates are 25.8% and 8.9%. The cited top three contain useful information but do not reliably identify the three factors with the strongest measured influence under necessity or sufficiency. The framework provides a black-box reliability check for explanations used in agent oversight while remaining scoped to individual LLM decisions.

Large language models (LLMs) are increasingly embedded in organizational work, yet their errors often pass human review. Prior research locates such failures in users' capability to review LLM output or their engagement in doing so. We develop an alternative, retrieval-based account of human oversight and posit that error detection is more effective when oversight-relevant information is accessible to users at the moment of review. Across two randomized lab-in-the-field experiments with 640 customer-facing employees, we show that self-generated explanations improve error detection and strengthen recall of verification-relevant reasoning, while cues that reactivate such reasoning help sustain detection under repeated LLM use. Theoretically, we identify information retrievability as a distinct precondition for effective oversight and specify generative encoding and cue-supported reactivation as mechanisms that build and sustain it. Practically, lightweight onboarding self-explanations and daily retrieval cues can make human oversight more resilient as LLM use becomes routine.

Large language models (LLMs) are used as post hoc explainers of sequential decision-making policies, producing natural-language explanations of why an action was chosen. However, LLMs often generate plausible but incorrect statements, and no existing approach systematically tests whether such explanations are faithful to the underlying environment. Two classic software testing challenges stand in the way: there is no oracle for the correctness of an explanation, and the test inputs, natural language queries about a policy's behavior, lack the structure needed for systematic test case generation. We address both. Probabilistic model checking provides the test oracle, computing exact reference results against which LLM answers are graded automatically. A taxonomy of post hoc query categories structures the input space around the environment-level facts from which policy explanations are composed; test cases generated from it are prioritized by question-specific diagnostic difficulty scores. Across seven MDP environments, the testing separates three open-weight LLMs: a reasoning model passes 85% of test cases, a mid-size model 70%, and a 1B model falls below the random baseline, while prioritization surfaces significantly harder cases than random selection. Our results indicate how trustworthy LLM-generated explanations are in model-free settings, where the same LLMs are used but no oracle exists to verify them.

To effectively enable people to understand new topics, explanations should be tailored to their backgrounds and abilities. Prompting alone has been shown to be insufficient for creating such explanations, and other computational methods are missing so far. Therefore, this paper investigates whether LLMs can be steered to generate explanations that are tailored to a specific group of people. To this end, we propose an approach that first identifies group-specific attributes in terms of explanatory style and knowledge of a specific target group. Building on activation engineering, it then computes attribute-based steering vectors and adds them to the internal activations of an LLM during inference to enable a fine-grained steering. In our experiments, we assess the steering effectiveness in terms of specificity and factuality of the generated explanations. Additionally, we evaluate the explanations in a study with human experts from different target groups. Compared to prompting and state-of-the-art steering baselines, our approach tailors the explanations significantly better to the target group while maintaining the best specificity-factuality balance.

Large-scale vision--language--action (VLA) policies have advanced generalist robot control, yet most remain stimulus-to-action black boxes: actions are exposed, but their explanatory state is not. They provide no native account of three explanatory signals: task progress, the next semantic transition, or local command reliability. Prior work shows that progress and event structure aid long-horizon control and that uncertainty supports monitoring; however, such capabilities are typically added or extracted only after action pretraining. The field therefore lacks a VLA foundation model whose explanatory state is jointly pretrained with control. Drawing on biological sensorimotor organization, in which outcome-sensitive, event-segmented, and probabilistic predictions structure behavior, we introduce LM-X. LM-X learns three directly supervised online signals: return-to-go (RTG) estimates visible progress and state quality; event-to-go (ETG) predicts the action sequence to the next semantic event; and heteroscedastic action-flow variance reports local command reliability. RTG conditions ETG and both condition action generation; uncertainty is estimated inside the action expert, making explanation part of control rather than a post-hoc description. We pretrain LM-X on more than 20,000 hours of heterogeneous real-robot trajectories, including over 1,000 hours of failed rollouts. A controlled gate favors joint over post-hoc training. LM-X achieves 74.1% success on 50 randomized-hard RoboTwin2.0 tasks and 73.5% on seven real-robot tasks, compared with 55.4% and 50.7% for GR00T N1.7. Its signals track progress and regression, anticipate event-scale motion, detect high-error actions, and provide advance failure warning. These results establish LM-X as an explainable VLA foundation model that couples transparent predictive state with stronger generalist control.

LLM-as-a-Judge, which leverages a large language model to evaluate natural language generated by another AI application or model, has become a standard, scalable approach for accelerating and extending costly human evaluation. Yet most work treats a judge as a static artifact, evaluating it once at construction or against a fixed benchmark. We argue instead that an LLM judge operating in a deployed system is better understood as having a lifecycle. It must be built, trained, deployed, and continuously maintained as the surrounding data evolves, and each phase poses distinct technical and operational challenges. We present such a lifecycle for the LLM judges that evaluate recommendation explanations at Netflix. Everything we report comes out of a series of controlled online member-facing experiments, in which our pipeline generated and the judges assessed hundreds of thousands of distinct show-level explanations per week across a changing catalog. Our framework has four phases. (I) Birth defines the evaluation criteria and builds curated benchmark datasets with human labels and rationales. (II) Training refines the judges' rubrics via Reasoning-Aligned Rubric Tuning (RART), which uses a meta-judge over reasoning output as the learning signal. (III) Deployment puts one judge in two online roles, quality gating and reflective generation. (IV) Monitoring runs a continuous Human-in-the-Loop (HITL) alignment process that detects drift and triggers re-tuning behind a human review gate. We report results from a five-week online A/B test over tens of millions of members on the Netflix mobile app, in which judge-aligned explanations shifted member viewing toward novel content (previously unwatched) and increased successful browse-to-play sessions relative to a no-explanation control, with no quality-related escalations.

Sparse autoencoders (SAEs) are proposed to extract numerous features from large language model (LLM) representations, yet explaining these features still relies primarily on external observation. This reliance leads to superficial explanations inferred from observed model behavior and computational inefficiency from collecting such behavioral evidence at scale. We introduce SAEVerbalizer, a framework that injects SAE decoder directions into an LLM's representations and fine-tunes the LLM's downstream layers to generate natural-language explanations of the injected features. Once trained, the resulting verbalizer explains SAE features directly from decoder directions, addressing both limitations. Our experiments show that the learned verbalization capability generalizes to unseen features, transfers across separately trained SAE dictionaries, and, with a lightweight adapter, extends to SAE features from different LLMs. Intervention experiments show that injecting multiple directions yields an explanation combining their meanings, while reversing individual directions produces corresponding meaning shifts.

Prior work on LLM behavior under anomalous conditions asks whether a model notices anomalies. We ask a narrower question: once a model sits in a workflow with a low, controllable failure rate, does its explanatory engagement - length, specificity, self-reported confidence - change as failure grows asymptotically rarer? We built a local, zero-cost harness on three open-weight models (qwen3:8b, llama3.1:8b, mistral:7b) running a repeated tool-call task where one call fails at probability p, swept across eight rates from 0.2 to 0.0001, under five elicitation conditions from immediate prompting to none. We hypothesized a rise in engagement as failures grew rarer, then a collapse near a detectability threshold. Pooled across conditions this appeared false: length fell in a flat, monotonic pattern. Splitting by condition overturned that. Under immediate_forced, where the model must explain every failure instantly, the predicted rise is confirmed but followed by a plateau, not a collapse: length peaks at 28.4 words at p=0.05, settles to 17.4-19.0 words at the rarest rates, and confidence rises unevenly from about 53% to the 70s-90s. Under grouped_runs, explanation batched to run-end, no collapse appears. Under passive_unprompted, aggregate magnitude is a floor artifact, but a recovered logging gap revealed real, model-specific self-monitoring: llama3.1:8b volunteers structured confidence reports unprompted, sometimes eroding its own confidence as trials accumulate; the other two do so only once, as boilerplate. Elicitation structure is a first-class moderator of collapse observability. A companion guaranteed-failure run (72 cells, backfilling rates where random sampling gave zero real failures) shows models differ in whether they recognize an anomaly, distinct from engagement once recognized. Limitation: discrete rate points cannot capture behavior between them, a direction for future work.

The performance of large language model (LLM)-based multi-agent systems (MAS) largely depends on effective communication topologies. Existing topology generation methods, however, typically learn communication topologies through black-box optimization driven solely by task-level rewards. While effective, such optimization provides little insight into why particular communication edges are selected, making it difficult to identify the critical communication subgraphs responsible for successful collaboration. To address this limitation, we propose E2-Explainer, a model-agnostic framework for providing interpretable explanations of communication topologies produced by arbitrary topology generators. Specifically, we formulate topology explanation as a causal attribution problem that identifies compact communication subgraphs supported by edge-level evidence of task preservation. We obtain this evidence with a Granger-style objective that measures how masking each communication channel changes the task outcome and the stability of the final response. The resulting budgeted subgraphs are then distilled into an amortized explainer, enabling efficient post-hoc explanation without repeated edge-level evaluations at deployment. Extensive experiments on multiple reasoning and coding benchmarks demonstrate that E2-Explainer identifies critical communication subgraphs that preserve successful collaboration. These subgraphs can also be executed directly to prune redundant communication edges, substantially reducing communication costs while maintaining competitive task performance.

Language model classifiers with explanations are used for moderation, routing, topic triage, and low-resource annotation. We study black-box auditing when the defender has only clean calibration data without trigger information but can ask the classifier for a label plus a short rationale or quoted evidence. We introduce Groundedness Drift, a lightweight score measuring whether the answer summary remains grounded in the input. Across two 7B backbones, five datasets, and four common non-adaptive OpenBackdoor-style attack families, Groundedness Drift achieves higher AUROC and lower residual target ASR than every compared detector in all cases at a nominal 5% clean-FPR budget. We then evaluate Unsupported Groundedness, a multi-probe escalation for explanation-camouflage stress cases. Unsupported Groundedness improves signals but does not close the adaptive gap.

Class activation mapping (CAM) is one of the most widely used visual explanation families in explainable artificial intelligence. Its purpose is intuitive: it converts internal model evidence into a heatmap that highlights the image regions, convolutional channels, tokens, or patches that support a target class or concept. Since the first CAM formulation in 2016, the field has moved far beyond global-average-pooled CNN classifiers. CAM-style methods now include gradient-based post-hoc explanations, gradient-free score and ablation methods, high-resolution upscaling, weakly supervised localization and segmentation, transformer token attribution, causal and debiasing methods, and foundation-model-era approaches that use CLIP, DINO, SAM, or feature-distribution comparisons. This review synthesizes a strict corpus of 57 method-centered papers published from 2016 onward. The paper develops a taxonomy that separates methods by attribution mechanism, architectural dependence, and evaluation objective. It then reviews gradient-based CAMs, recent and hybrid CAM-style methods, and model-based or architecture-aware methods. Across the corpus, the main trend is clear: the field is shifting from explaining one class score in one low-resolution CNN layer toward comparative, multi-layer, probabilistic, token-aware, and foundation-model-aware explanations. At the same time, evaluation remains fragmented. Faithfulness, localization, robustness, computational cost, and human trust are often measured with different protocols. The review therefore emphasizes not only what each method contributes, but also which gap it leaves open and which later methods attempt to close that gap.

Machine learning pipelines commonly flatten relational data into single-table representations, discarding structural constraints. Widely used Shapley value-based feature attributions then rely on feature independence, evaluating the model on combinations that could never arise in the underlying data, producing misleading explanations. We propose RelShap, a framework that incorporates relational constraints and data provenance into Shapley value computation, restricting both background data and coalition evaluation to relationally valid configurations. The framework is estimator-agnostic and composes with Kernel SHAP, Monte Carlo, and Leverage SHAP without altering their sampling or weighting properties. Functional dependencies further induce equivalence classes over feature coalitions, which RelShap exploits to reduce runtime without changing Shapley values; we provide a combinatorial characterization of the expected speedup. Experiments across multiple datasets, models, and estimators show that RelShap produces explanations that are more faithful to the data-generating process, correctly identifying the dominant feature in controlled settings where existing methods, including Conditional SHAP and ManifoldShap, do not. Our code is available at: https://github.com/duneag2/relshap.

Explainability is essential for deploying deep learning models in high-stakes medical applications. Existing explainability methods for volumetric imaging predominantly operate in voxel space, overlooking the structured representations introduced by recent advances in 3D scene modeling. We present COGENT (Counterfactual Gaussian Explanations), a framework that generates counterfactual explanations directly in the parameter space of Gaussian-based volumetric representations. Built upon MedGS and the Sybil lung cancer risk prediction model, COGENT optimizes selected Gaussian primitives through a differentiable rendering pipeline, enabling gradients from the downstream predictor to identify representation components that most influence model decisions. Unlike conventional pixel- or voxel-level attribution methods, our approach formulates explainability as a counterfactual optimization problem over an explicit 3D scene representation, producing sparse and spatially localized explanations while preserving anatomical consistency. We evaluate COGENT on lung CT scans using quantitative comparisons with existing explainability methods together with qualitative analysis by medical experts. The results demonstrate that representation-space counterfactual optimization provides clinically meaningful explanations while offering a new perspective on interpreting volumetric deep learning models.

Artificial Intelligence (AI)-based prospective anomaly detection methods are increasingly deployed in high-dimensional and nonlinear settings. Among these approaches, AI-based statistical process monitoring (SPM) is widely used, providing a structured framework for prospective monitoring. Once an anomaly is detected, a diagnosis method is needed to identify the features driving the flagged observation away from normal behaviour. Traditional SPM diagnosis methods are typically designed for specific detection models and cannot be directly applied to AI-based methods. Model-agnostic explainable AI (XAI) offers a general framework for feature relevance explanation. However, existing methods suffer from scalability limitations or assign relevance to noise features, reducing diagnosis accuracy. We propose a scalable, baseline-referenced diagnosis method that uses both the anomalous observation and normal baseline information. We provide mathematical guarantees that under a mean-shift anomaly setting, the proposed method achieves higher faithfulness in detecting the features causing the anomaly compared to LIME. Simulation studies and a real-world case study validate the effectiveness of the proposed method and show that it generates more faithful and accurate diagnosis results for AI-based prospective anomaly detection methods.

In the operation of machine learning models, model update is a fundamental process that requires careful consideration of its impact on downstream decision-making. Particularly when operating explainable models, changes in explanations resulting from model updates can lead to detrimental outcomes for users. Decision trees, due to their high transparency, are frequently employed in risk-sensitive decision-making and serve as a prominent example in which the aforementioned issue is evident. However, existing research addressing similar issues has focused on explanations based on feature contributions, and thus cannot handle explanations derived from tree structures. Therefore, this paper proposes the Backward Compatibility Loss in Tree-based eXplanations (BCLTX), a loss metric that suppresses changes in decision tree explanations before and after updates. Furthermore, we design CART with Backward Compatibility in Tree-based eXplanations (CART-BCTX), a lightweight algorithm that improves upon CART for the decision tree update problem under BCLTX. Experimental results using 10 real-world datasets, including both classification and regression tasks, show that CART-BCTX achieves favorable trade-offs between prediction performances and BCLTX values, with comparable computation times to CART, regardless of the task.

Visual counterfactual explanations aim to change classifier decisions through realistic and localized edits while preserving decision-irrelevant content. Existing DDPM-based methods typically perform classifier-guided editing along a long reverse denoising trajectory. The changing noise levels make semantic editability and spatial control difficult to balance, and the editable state is noisy, whereas the target classifier is trained on clean images. As a result, these methods require either costly recursive denoising or low-quality one-step estimates to obtain classifier-facing clean images. We propose FiRe, a Fixed-noise Refinement framework for visual counterfactual explanations. Rather than following a reverse denoising trajectory, FiRe maps the input to a fixed noise level and iteratively refines the noisy state at that level. To provide clean images for classifier guidance, FiRe first adapts Pixel Mean Flow to visual counterfactual explanation, enabling direct clean-image prediction from noisy states. To make fixed-noise refinement produce minimal and localized counterfactual edits, FiRe introduces three FiRe-specific controls: a dynamic dual-mask strategy, adaptive guidance, and early stopping, which determine where edits accumulate, which changes become visible, and when refinement stops. Experiments on five tasks across three datasets show that, compared with the strongest recent baseline, FiRe achieves about 3$\times$ faster online inference and 8$\times$ fewer FLOPs while obtaining comparable or state-of-the-art counterfactual quality.

Credit scoring increasingly relies on models whose decision logic cannot be read off their parameters, in tension with supervisory expectations that adverse decisions be explainable. A common proposal closes that gap with a language model: compute feature attributions, hand them to an LLM, and let it write the rationale. We build such a system end to end and test whether the second half of the promise holds. The predictive component is a multi-scale stacking ensemble fusing four differently regularised gradient-boosting learners with a residual network through a neural meta-learner trained on out-of-fold predictions. On a public 32,581-application credit dataset it reaches test ROC-AUC 0.9539 (95% CI [0.9462, 0.9616]) and PR-AUC 0.9137, beating the best single model by Delta-AUC = 0.0143 (p = 0.016 under a conservative independence assumption). Our central finding is asymmetric. The ranking gain is real but operationally small: at the F1-optimal threshold the ensemble avoids only six additional missed defaults out of 1,422 against a tuned random forest, cutting cost-weighted loss by under 2%. The narrative layer fails in a way prompt engineering alone does not fix. In an audited case the model named three factors as risk-increasing that the supplied attributions scored as risk-reducing, omitted the dominant driver, and introduced a feature never given to it. We trace this to properties we measure rather than assume: SHAP and LIME agree on which features matter (overlap@10 = 0.80) but not on their order (tau = 0.43, p = 0.18), and the attribution sign for the model's most sensitive input is near a coin flip across applicants (modal-sign share 0.53). Calibration (ECS = 0.117) and perturbation stability (DPD = 0.078) both fall short of our own thresholds. Constrained prompting is necessary but not sufficient: grounding must be verified after generation, not assumed.

This paper addresses the limitations of Explainable Artificial Intelligence (XAI) with respect to insufficient evaluation. They are illustrated through the DetoxAI image recognition system for bias detection and concept unlearning. Then, an example of a human-grounded evaluation of methods for explaining image classification is presented. The paper further explores methods for adapting explanations to evolving data streams with concept drift. Experiences with adapting counterfactuals for this problem are discussed. Finally it is related to the challenges of tracking the co-evolution of data, models, and explanations.\footnote{This paper has been accepted for a publication in J.Nalepa (ed) Explainable AI in Space. Proceedings of EASi 2026 Workshop at IJCAI-ECAI 2026 Bremen, Springer CCIS vol 3107 (2016).}

Large language models (LLMs) have been shown to exhibit strong Python preferences when generating project-level code, but there is currently no systematic way to measure this behaviour across new models. To bridge this gap, we introduce LangChoiceBench, a project-level code-generation benchmark for measuring Python preference, recommendation-implementation consistency, and language diversity. LangChoiceBench covers 28 projects across seven software areas where Python is often a poor default. We evaluate 25 diverse LLMs and find that Python remains heavily over-selected, recommendation-implementation consistency is low, and smaller open-weight models generally show stronger Python preference and lower language diversity. We further analyse 9,826 reasoning traces and find that most Python choices are automatic or driven primarily by ease, rather than explicit consideration of project requirements. In a smaller but important set of cases, models fabricate contextual support for choosing Python - a failure mode we call phantom evidence - or produce code that contradicts the language selected in their own reasoning.

An information theory-based framework is proposed in attempt to explain insistence on sameness in autism as an instance of a general behavior pattern in which an individual tries to reduce surprise and uncertainty. It offers a new definition of autism as an impairment in which cognitive functions are restricted to discrimination, memorization and prediction of tangible properties of the environment. An analogy between insistence on sameness and constrained minimization of the entropy metric is observed and examined for a set of assumptions that describe cognitive limitations of a person with autism. The metric is given by the formula $D_H(R, M) = H(R|M) + H(M|R)$, where $R$ represents sequences of random stimuli, $M$ is a memory that stores and retrieves them, and where $H(.|.)$ denotes their conditional entropies interpreted as surprise and uncertainty, respectively. It is first inferred that to minimize the metric an individual can learn about $R$ (and store that knowledge in $M$) or can restrict $R$ to the already known $M$. Then, it is concluded that insistence on sameness is a manifestation of the latter. Moreover, it is shown that the proposed framework: (1) Helps to quantify the concepts of surprise, uncertainty, sensory overload and deprivation, anxiety, comfort zone, disappointment, disorientation, pedantry, rigidness, observance or aberrant precision. (2) Leads to a list of guidelines for learning therapies and daily care routines, and allows them to be defined as optimization algorithms and implemented as programs for robotic live-in caregivers. (3) Can be validated with the help of a Turing test-like approach that requires no experiments involving individuals with autism. The framework-if positively validated-will provide formal foundations and design guidelines for therapies aimed at improving self-reliance of individuals with autism in basic activities of daily living.

Explaining the predictions of neural networks is a central challenge in trustworthy AI. Existing explanation methods, such as those based on feature attribution or minimal sufficient sets, typically treat input features as independent, which can yield misleading explanations when inputs exhibit structured dependencies. We address this by formalizing explanations as Halpern-Pearl (HP) actual causes, modeling input dependencies using Boolean Structural Causal Models (SCMs). We compute HP causes by applying bound propagation and branch-and-bound techniques, while providing formal guarantees of completeness and minimality. Our experiments show that we substantially outperform brute-force and ILP baselines in scalability, and outperform heuristic search as graph size grows, computing all minimal actual causes on instances with search spaces of up to $2.3\times10^{13}$ candidate (cause, contingency) pairs, on SCMs with up to 28 nodes, within a 180s per-instance budget. In a case study, we further show that ignoring input dependencies inflates the number of reported causes, 14.9% of which are spurious under our SCM.

Statistical reasoning is multidimensional, yet evaluations of large language models (LLMs) typically emphasize response accuracy while overlooking how models construct and communicate statistical explanations. This study demonstrates the value of a multidimensional evaluation by combining response accuracy, response behavior, structural topic modeling, and lexical similarity analysis. The framework is applied to explanations generated by 15 current-generation LLMs responding to 90 questions drawn from four statistics examinations spanning high school, undergraduate, and graduate levels. Accuracy varied substantially across models, ranging from 55% to 78%. In contrast, structural topic modeling revealed a common conceptual organization of statistical reasoning across all models, while lexical similarity analysis identified modest but consistent vendor-specific differences in explanatory style. Models developed by the same vendor (e.g. Anthropic, OpenAI) produced explanations that were slightly more similar than models from different vendors. These findings demonstrate that statistical reasoning in contemporary LLMs cannot be characterized by accuracy alone and illustrate how complementary analyses of response behavior and model-generated explanations provide a more comprehensive evaluation of statistical reasoning in generative AI.

When algorithms make or influence consequential decisions---about loan eligibility, hiring, or healthcare---EU law grants affected individuals a Right to Explanation. Yet whether (and how) Explainable AI (XAI) can satisfy this right in practice remains poorly understood, with direct implications for individuals' ability to contest automated decisions that affect their lives. This paper presents a systematic literature review of XAI in the context of the EU Right to Explanation, with particular focus on Art. 15(1)(h) GDPR, Art. 86 AI Act (AIA), and related instruments. We consider papers published from 2024 onwards, as the final version of the AIA was published in July 2024---with Art. 86 being added late. From 2643 initial records identified by a deliberately broad search, we review 57 full texts, of which only 19 papers demonstrate substantive integration of both legal and technical perspectives, showing gaps in the interdisciplinary synthesis of the current regulatory framework. We document three problematic patterns across the corpus: Most misidentify the GDPR legal basis; few engage with the CJEU's Dun & Bradstreet judgment (likely due to publication timing); and the distinction between explanation form (governed by addressee) and content (governed by legal purpose) is often conflated. We conceptualize this as the Addressee/Purpose Framework, propose a four-phase blueprint for operationalization, and identify six concrete open research questions. Without further progress, the Right to Explanation risks remaining a formal obligation without a technically realizable path to compliance.

Simulating human-like Theory of Mind (ToM) has been a longstanding problem in natural language processing (NLP). To address this, existing works introduce a reasoning step of event hiding (a.k.a. perspective-taking), where events unknown to a character are removed before question answering. However, resorting to event hiding for ToM reasoning presents a performance degradation issue due to the strict output format constraints involved in event hiding. To mitigate this issue, we propose generating perspective-taking outputs as free-form explanations without event hiding, but this poses a notable yet underexplored challenge: LLMs need to inhibit responses to events unknown to characters, because the absence of event hiding exposes LLMs to these events throughout reasoning. To address this challenge, we hypothesize and empirically verify that LLMs can achieve such inhibition if a character's lack of knowledge about events is made explicit during reasoning. Based on this finding, we introduce PICTURE, a new prompting method that enables LLMs to generate a character's lack of knowledge within free-form Chain-of-Thought (CoT). Experimental results show that PICTURE outperforms existing prompting methods by an average of 7.3% on false-belief tasks.

Vision-language models (VLMs) process image patches and text tokens in a shared residual stream, but the local geometry through which the two modalities interact remains poorly understood. Most interpretability methods identify global linear directions, which may miss representations that are globally high-dimensional but locally low-dimensional. We introduce LENS (Local Explanation of Neighborhood Subspaces), a method that decomposes VLM activations into local low-rank Gaussian neighborhoods using a Mixture of Factor Analyzers. Applied to LLaVA-1.5-7B and Qwen3-VL-8B, LENS reveals distinct depth-dependent fusion trajectories consistent with each model's fusion mechanism: LLaVA progressively mixes modalities at later layers, whereas Qwen3-VL mixes them early, partially re-segregates them, and recombines them near the output. An automated multimodal labeling pipeline assigns concise semantic descriptions to these neighborhoods. Interpolating activations toward neighborhood centroids causally redirects generation within and across modalities and outperforms difference-in-means and VL-SAE in most evaluated conditions; in one LLaVA vision-to-vision setting, MFA achieves 5.7 times the VL-SAE score. Human evaluation finds MFA steering competitive with prompting and substantially stronger than the other intervention baselines. Finally, the MFA coefficient space improves Qwen3-VL image-to-rendered-text retrieval at the deepest evaluated layer from 14.9% to 48.6% R@1. Ablations show that the reported fusion trajectories are stable across component counts, local ranks, and modality-purity thresholds. These results support local geometric neighborhoods as useful interpretable and causal units for analyzing cross-modal representations in the evaluated VLMs.

Language models are widely used for generating and otherwise processing code (e.g., identifying code hallucinations, possible inputs, or predicting outputs); however, LLMs can make mistakes, which can be serious. One key issue is that models are trained on (still) largely human-written, and thus imperfect, code; it's not easy to find sufficiently large code corpora that are entirely free of bugs. Thus, other inference-time ways of reducing LLM errors, without additional training, are desirable. "Reasoning" or "thinking" modes, exposed as a togglable feature by hybrid reasoning models, do reduce errors; however, reasoning consumes additional resources. This paper asks if better performance can be achieved without always incurring the cost of reasoning. Human students of programming learn to avoid mistakes by (a) identifying them, (b) reflecting upon the cognitive lapses that led to them (essentially, "thinking through" the errors), (c) inferring general rules or lessons from these reflections, and (d) internalizing these lessons into rules. In tutorial sessions with an instructor, this is a common Socratic interaction. Examples of such internalizable rules might include the nugget "Before coding, restate the requirements to clarify them." Inspired by this process, this paper describes an approach where we first identify examples in which "thinking mode" in a (low-resource) LLM avoids errors. These errors, and their avoidance via "thinking" in the same LLM, are then examined by a bigger LLM to generate summary explanations; these are then summarized by a large LLM into brief advisory prompts. This approach works on many modest-sized models; in some cases, the "advisory prompts" thus learned can also be gainfully transferred to other models. We also present investigations into the nature of coding errors that language models make, and a characterization of when this approach can be helpful.

Datalog rules are often used to define ontologies over Knowledge Graphs. Rule reasoners routinely optimise such ontologies by rewriting their rules into a form that can be evaluated more efficiently. These transformations preserve the entailed facts, but not the structure of the underlying derivations. A proof tree under the rewritten rules explains why a fact holds, but does not readily yield an explanation in terms of the original rules. We study the problem of constructing, from a proof of entailment under the rewritten rules, a proof under the original ones: we establish its computational complexity and identify two practically relevant languages for specifying proof transformations.

The rapid adoption of deep learning models in high-risk domains has intensified the need for trustworthy Explainable Artificial Intelligence (XAI). However, objectively evaluating explanation fidelity and aligning XAI metrics with human-centered understanding remain critical open challenges. In this work, we propose a model-agnostic metric, the EPC score, which is an extension of the Explainability-Performance Coefficient (EPC), that quantifies explanation quality by explicitly balancing the trade-off between feature selection sparsity and preserved model performance. Through an empirical validation across tabular, text, and image modalities, we show that the EPC score effectively uncovers operational dependencies among network activations, data dimensionality, and explainer performance. Furthermore, we validate the EPC score against independent human-based explanations, proving that higher EPC scores strongly align with human lexical sentiment judgments and spatial visual annotations.

Dimensionality Reduction (DR) is a fundamental tool for high-dimensional data exploration, reducing the complexity of latent spaces of machine learning models, and assisting in the explanation of complex opaque models. However, non-linear DR techniques often function as opaque transformations themselves, making it challenging to understand how individual features influence instance positioning in the reduced space. This lack of transparency complicates the analysis and interpretation of structural patterns, hindering the ability to reason about the organization of high-dimensional data based on the projected layout. In order to address this challenge, dimensionality reduction explanation methods have shown promise in improving the understanding of the observed groups and cluster structures. Unfortunately, existing DR explanation approaches tend to suffer from limitations such as multiple attributions per feature and restricted applicability to specific dimensionality reduction methods, which hinder their use. In this work, we propose FADEx, a novel local per-instance feature attribution method that leverages local linear approximation via first-order Taylor expansion and Singular Value Decomposition to provide explanations. FADEx computes the local linear models via weighted least squares, eliminating the need for out-of-sample data mapping, making it agnostic to the DR method, while simultaneously providing local feature attributions and distortion analysis. Through qualitative and quantitative evaluations, comparisons with existing methods, and case studies, we demonstrate FADEx's effectiveness and versatility in providing explanations and analytical resources for analyzing the behavior of DR methods. The results indicate FADEx yields robust and reliable explanations, outperforming existing approaches in several aspects.

A gradient-based GNN explainer given a molecule with two chemically equivalent nitro groups assigns them attribution scores that are equal to the last bit. It cannot do otherwise: message passing is exactly permutation equivariant, so any automorphism of the input leaves every attribution invariant. Yet the standard report, the top-k edges, names one of the two, and which one is settled by the order of an array. We show this is a structural obstruction rather than an implementation slip. When no minimal valid explanation is fixed by the input's automorphism group, no rule can be single-valued, minimal and symmetry-respecting at once. For the exact-k reports used in practice we give a parameter-free criterion, mechanised in Lean 4 with no axiom dependencies, that decides from the graph alone whether every score-optimal report of that size must split an orbit. Across 21298 instance-budget decisions the criterion agrees with a mechanical model-equivalence check without exception, and no severing case we found admitted a neutral alternative. The obstruction is common. Nontrivial automorphisms occur in 93.4% of Mutagenicity, the dataset the seminal explainability papers use, so the measure-zero dismissal of symmetric inputs, sound on the continuous domains it was made for, collapses here. At the sparsity budget those papers report, 24.0% of molecules with two interchangeable nitro groups (6 of 25) surface exactly one of them, every one arbitrary under mechanical verification. A model's blindness also manufactures symmetry: every MUTAG molecule contains atoms chemistry separates and the network provably cannot, and a matched control shows the resolution is set by what the model reads rather than how it is parameterised. Reporting orbits removes the arbitrariness at 0.11 ms and 0.43 extra edges per graph.

Security operations centers rely on anomaly detection systems to flag suspicious events. Feature-level explanations for anomaly detectors offer limited value for operational investigations. To effectively handle alerts, analysts need to know contextual relationships and need actionable understanding of the entities involved. This paper introduces an event-centric detector-agnostic approach for explaining cybersecurity alerts in small- to medium-sized enterprise networks. We present (EC)2, a multi-agent framework that performs structured, hypothesis-driven investigation to provide explanations grounded in verifiable evidence. Evaluation results show that the proposed framework improves post-detection analysis by generating operationally meaningful explanations, which also enhance event classification accuracy.