Research Papers

Agreement with human judgments is a common proxy for evaluating the alignment of large language models (LLMs). Yet agreement in final labels does not show that human annotators and models rely on the same moral grounds. Two agents may reach the same judgment while appealing to different principles, contextual assumptions, or interpretations of the situation. We test this distinction using a curated 500-item ETHICS-derived benchmark spanning five domains of moral judgment, with new human annotator and LLM annotations of both final labels and supporting rationales. Across frontier and open model families, agreement with human annotator majority labels is often high. However, rationale-level analysis reveals systematic divergence in the moral grounds expressed by human annotators and models. In particular, models redistribute attention across categories such as harm, respect, promise-keeping, justice, desert, and excuse relevance, even when their final labels match the human annotator majority. Our results show that agreement should not be treated as equivalent to alignment. Label-based evaluation can therefore be misleadingly reassuring unless complemented by analysis of the reasons, principles, and moral priorities expressed in model judgments.

Evaluating RE Practices for Explainability: Synthesizing Insights from Daimler Truck into an Explainable RE Framework Proposal

Umm-e- Habiba, Lucas Mauser, Jonas Fritzsch +2

Explainability has emerged as a critical requirement for AI-based systems, particularly in safety-critical and regulated domains. Although prior research has proposed frameworks, patterns, and user-centered approaches to support explainability, there is limited empirical understanding of how existing Requirements Engineering (RE) practices support explainability requirements across the RE lifecycle, especially in an industrial context. This paper reports early findings from an ongoing industry-based study investigating how explainability requirements are elicited, specified, and validated using established RE techniques. We conducted a multi-phase qualitative study with eight practitioners at Daimler Truck, employing think-aloud protocols and moderated group discussions across requirements elicitation, specification, and validation steps. Our preliminary analysis reveals recurring challenges across all steps, including conceptual ambiguity during elicitation, limited testability and expressiveness during specification, and fragmented validation due to vague criteria and regulatory uncertainty. These findings indicate that current RE practices provide limited support to systematically address explainability requirements. The paper contributes empirical insights into step-specific and cross-cutting challenges and outlines a research vision toward developing an empirically grounded RE framework for explainable AI-based systems.

Explainable machine learning (XML) pipelines applied to composite mental health outcomes can produce apparently-robust, cross-population-stable risk hierarchies that are largely artefacts of how the outcome was constructed. We demonstrate this using an ElasticNet pipeline applied to 886 medical students at the University of Lausanne (primary cohort, 2022), validated across 2,580 longitudinal observations at three time points and 701 non-medical students from eight faculties; all three datasets share identical instruments. The pipeline produces a hierarchy in which trait anxiety and health satisfaction dominate wherever the outcome is measured, with Kendall $τ= 1.0$ for the top-two positions across all five evaluation sets and consistent transfer performance ($R^2$: 0.41-0.49). Two residualization experiments, which isolate shared variance between correlated variables via regression, reveal the mechanism: when trait anxiety (STAI-T) is residualized against the co-included depression subscale (CES-D, $r = 0.72$), model $R^2$ drops from 0.41 to 0.16 and STAI-T falls from rank 1 to rank 6; when burnout subscales are residualized against CES-D, $R^2$ collapses to 0.016. Prediction intervals average 35.4 units on a 0-100 scale (2.4 outcome standard deviations), independently ruling out individual-level deployment. The residualization protocol is the paper's transferable contribution: any XAI study combining correlated predictor and outcome constructs should apply this check before interpreting apparent stability as a finding.

Algorithmic fairness audits can detect disparities, but they do not determine when those disparities warrant intervention. Deployment decisions also depend on the reliability of the evidence, subgroup support, and deployment context. Existing fairness methods quantify disparities and uncertainty, yet provide limited guidance for translating accumulated evidence into action. We introduce Statistical Actionability, a statistical construct that recasts fairness deployment as an evidence-based decision problem. The framework integrates fairness evidence regarding disparity magnitude, statistical reliability, subgroup adequacy, and deployment context, and maps the resulting evidence state to one of four recommendations: mitigate, collect more data, monitor, or take no immediate action. In controlled simulations, Statistical Actionability achieved the lowest decision cost among representative baselines, reducing average decision cost by 19.2% relative to gap-based intervention while simultaneously reducing both false alarms and missed bias. A calibrated deployment rule generalized across heterogeneous statistical environments, remaining within 2% of the target oracle in four of five transportability regimes. Analyses of benchmark fairness audits further demonstrated that the framework distinguished audits with similar observed fairness gaps but different levels of uncertainty and subgroup support, yielding interpretable deployment recommendations. Statistical Actionability therefore establishes a statistical decision layer between fairness evaluation and deployment intervention, enabling responsible AI systems to act on accumulated evidence rather than disparity magnitude alone.

Large language models (LLMs) are increasingly integrated into financial advisory systems, yet their role in reproducing religious bias remains underexamined. This study provides systematic mixed-methods evidence of such bias across three LLMs (ChatGPT, Gemini, and Grok) using 432 simulated advisor-client interactions spanning 16 religious identity pairings (Christian, Muslim, Hindu, and non-religious) and three core household financial decisions: stock investment, house purchase, and life insurance. Combining regression and reflexive thematic analyses, we identify structural biases across models and decision contexts and the discursive mechanisms through which they are linguistically enacted. Unbiased advice appeared in only 12-18% of cases. Gemini consistently produced more bias than Grok, while ChatGPT's outputs were statistically comparable to Grok's. Religiously symmetric advisor-client pairings almost always triggered explicit religious framing, and non-religious clients often received advisor-centered religious appeals. Qualitative findings show that bias is linguistically manifested through religious anchoring, uneven cultural signaling, and tone modulation, varying by model and financial scenario. Stock investment prompts produced more financially technical responses, whereas life insurance advice triggered stronger religious language. The study develops a dual-dimensional framework linking structural bias rooted in model training and design with discursive bias expressed through language, advancing understanding of algorithmic bias in LLM-generated financial advice. It also shows that such advice adapts linguistically to identity cues, revealing a managerial dilemma between personalization and neutrality. Finally, it highlights implications for businesses, financial institutions, and regulators seeking to ensure neutrality, cultural sensitivity, and trust in AI-mediated advice.

National AI strategies increasingly guide governance, workforce development, innovation, and competitiveness, but less is known about how they frame education as a sector with pedagogical, cultural, ethical, and implementation demands. This study develops and applies an Education-Centered AI Policy Framework to analyze Ghana's National Artificial Intelligence Strategy, 2025-2035. Using critical qualitative policy document analysis, we examined the strategy through six components: policy purpose, teacher agency and professional learning, curriculum and assessment, language and culture, responsible AI and learner protection, and participation and implementation governance. Findings show that Ghana's strategy is ambitious and timely, especially in its emphasis on AI literacy, youth skills, TVET, workforce readiness, rural outreach, local language data, inclusion, and responsible AI governance. However, the education agenda is stronger on national AI readiness than on school-level implementation. Teacher agency, pre-service teacher education, curriculum progression, assessment guidance, AI disclosure, multilingual pedagogy, culturally responsive AI use, child-centered safeguards, and participatory governance remain underdeveloped. We also identify document-level concerns about transparency and coherence, including apparent AI-styled visual content without visible disclosure and a mismatch between a vision and mission figure and its textual explanation. We argue that Ghana needs a sector-specific, education-centered AI policy and implementation pathway that connects workforce readiness with teacher preparation, curriculum reform, assessment redesign, learner protection, infrastructure, local language instruction, culturally responsive pedagogy, locally responsive AI tools, and participatory governance.

The emergence of metaverse platforms has created virtual economies that introduce new challenges related to fraud, bot activity, and illicit financial behavior. Despite growing interest in trustworthy metaverse analytics, existing datasets typically focus on user behavior, authentication, or financial transactions in isolation, limiting the development and reproducible evaluation of multimodal fraud detection methods. To address this gap, we present TSAI-MetaFraud, a multimodal, multi-task benchmark dataset for fraud analytics in virtual economies. TSAI-MetaFraud integrates behavioral, transactional, and graph-structured information while incorporating realistic fraud and automated bot scenarios. We define benchmark tasks including transaction fraud detection, cross-modal node classification, temporal link prediction, and weakly supervised fraud detection, and provide baseline evaluations using machine learning models and graph neural networks. By jointly capturing behavioral activity, financial interactions, and relational structure within a unified virtual economy, TSAI-MetaFraud provides a benchmark for advancing multimodal learning, graph mining, fraud analytics, and trustworthy AI in emerging metaverse ecosystems.

Pre-demolition assessment, the regulated audit process at the heart of urban mining, is an information process in which AI support must serve qualified auditors who remain accountable for the decisions taken. The relevant unit of value is not prediction accuracy alone, but the defensibility of the supported decisions: their legibility, plausibility, sourcing, and contestability. Explainable AI techniques and domain knowledge graphs each address parts of this requirement, and existing taxonomies have catalogued their integration. The literature is descriptively rich but structurally under-specified: what remains less developed is a structural account of why specific integrations produce artefacts neither resource can provide alone. This paper offers a complementarity-theoretic interpretation grounded in the IS resource-based tradition. We propose four consolidated KG-XAI integration modes (Lifting, Constraining, Typing, and Revising), each defined as a typed operation over XAI artefacts and knowledge-graph substrate structures. Each mode unlocks a distinct property of defensibility and contributes to the kind of regulatory artefact pre-demolition assessment demands. A fire-door example from the urban-mining process illustrates the modes using the W3C Linked Building Data stack and valuation extensions.

Concept-based explainable artificial intelligence (AI) can make model reasoning more human-understandable, but concept-level outputs are not automatically trustworthy. We introduce ConceptSMILE, a model-agnostic perturbation-based auditing framework for evaluating the reliability of concept-based explanations. Rather than replacing SMILE, ConceptSMILE extends its perturbation-based logic from feature- or region-level attribution to the auditing of human-understandable concept explanations. The framework perturbs input regions, measures concept-response shifts, applies locality weighting, and fits an XGBoost surrogate to approximate local concept behaviour. Reliability is assessed through attribution accuracy, surrogate fidelity, faithfulness, stability, and consistency. We evaluate ConceptSMILE on retinal fundus images by comparing MedSAM-derived visual concepts with VLM-based semantic concepts. Results show that reliability varies across concepts and pathways: MedSAM achieves stronger spatial attribution and the highest surrogate fidelity ($R^2 = 0.8503$, $R_w^2 = 0.8465$), while the VLM pathway shows stronger vessel faithfulness and stronger stability under selected artefact conditions. ConceptSMILE provides an independent audit layer for evaluating the trustworthiness of concept-based XAI.

Crime linkage analysis is used in many countries to identify series of offences that may have been committed by the same individual. In practice, specialist analysts manually search for behavioural and situational connections across large crime databases, an effort that is time-consuming, cognitively demanding, and can involve repeated exposure to disturbing material. To support this work, an Artificial Intelligence (AI)-enabled decision-support tool was co-developed with a UK law enforcement agency to assist analysts in identifying likely crime linkages. This paper reports an industrial evaluation of the crime-linkage tool. We conducted a mixed-methods usability study combining direct observation, eye-tracking, mouse-tracking, and surveys to examine how analysts engage with AI predictions and with the model features presented as explanations. Our findings show that analysts used the AI predictions selectively and frequently validated them against behavioural (non-AI) evidence, reflecting partial trust and an ongoing reliance on established analytical practices. We also found that analysts attended to the presented model features and valued their availability, while identifying opportunities to improve how explanations are presented and integrated into the workflow. Overall, our results highlight the need for AI-enabled decision-support tools to better integrate explanations and traditional analytical methods, and demonstrate the importance of in-situ evaluation for engineering usable and trustworthy AI in high-stakes settings.

Multimodal emotion and sentiment recognition is commonly addressed by early fusion, which concatenates modalities before classification, or late fusion, which combines independently trained unimodal predictors. Early fusion can be accurate but monolithic, while late fusion is modular but may lose cross-modal interactions. This paper revisits XAI-guided adaptive fusion (\xgaf), a tree-based mixture of unimodal and cross-modal experts whose sample-level weights are derived from TreeSHAP attribution magnitudes. We focus on the effect of SHAP attribution reduction when experts have unequal feature dimensionalities. In this setting, mean-abs and median-abs reductions can suppress high-dimensional cross-modal experts, whereas sum-abs reduction preserves total attribution mass. On MELD 7-class emotion recognition, sum-abs \xgaf{} nearly matches early fusion across three face-sequence aggregators; the Transformer variant reaches 0.5983 \wf{}, compared with 0.6018 for early fusion and 0.4598 for probability-average late fusion. McNemar testing shows no significant difference between sum-abs \xgaf{} and early fusion on MELD ($p=1.000$), while \xgaf{} remains significantly better than late fusion ($p<0.0001$). On CMU-MOSEI 3-class sentiment recognition, sum-abs \xgaf{} reaches 0.6519 \wf{}, slightly exceeding early fusion (0.6485) and late fusion (0.5696). Ablation studies show that the main gain comes from adding cross-modal experts, especially the trimodal expert, rather than from complex per-sample routing. Diagnostics further show that mean-abs and median-abs weights are nearly uniform, while sum-abs weights concentrate on the trimodal expert. Thus, the main contribution is a transparent empirical analysis of how SHAP reduction, expert dimensionality, and cross-modal expert design affect modular multimodal fusion.

The 2026 Singapore Consensus on Global AI Safety Research Priorities

Stephen Casper, Oskar Galeev, Yoshua Bengio +117

Frontier AI capabilities and autonomy are advancing rapidly. A growing number of real-world incidents make a trusted AI ecosystem essential to embracing AI with confidence. The 2026 Singapore Consensus is an outcome of the second International Scientific Exchange on AI Safety, bringing together over 100 contributors spanning 13 countries from frontier developers, government safety institutes, academia, and civil society. Building on the 2025 report, it presents a global understanding of technical AI safety research problems of top priority, now with a dedicated focus on societal resilience and on managing the risks of increasingly autonomous AI agents.