Papers

Rapid and reliable disaster mapping of impacted areas, damaged infrastructure, and affected populations is essential for emergency response and recovery. However, existing AI-based approaches often require extensive manual annotation, lack cross-hazard generalization, and rely on single-modal observations. To address these challenges, this paper proposes RAPIDMap, a rapid multi-agent pipeline for zero-shot interpretable disaster mapping from satellite and street-view imagery. The framework integrates four intelligent agents: Disaster Perception Agent (DPA), Image Restoration Agent (IRA), Damage Recognition Agent (DRA), and Disaster Mapping Agent (DMA). By combining remote sensing and street-view data, RAPIDMap eliminates the need for manual fine-tuning, generalizes across multiple disaster categories, and generates structured, map-ready disaster intelligence with recovery recommendations.

Immigrant-led nonprofit groups, particularly those operating in politically sensitive contexts, face exclusion from formal registries and digital platforms. This paper reports a three-phase mixed-methods study with Iranian immigrant nonprofit practitioners: 27 semi-structured interviews, a co-design session, and 7 evaluation and feedback interviews on a prototyped AI assistant, AMINA. Our findings highlight how legitimacy barriers, capacity gaps, and politically charged misinformation constrain nonprofit operations. We translate these insights into design goals for an inclusive nonprofit AI assistant: support for everyday group operations, recognition of informal nonprofit efforts, proactive countering of misinformation, and multilingual, accessible interaction. User evaluations show AMINAs potential to reduce reporting burdens and foster transparency through proactive reminders, and catalyze collaboration across dispersed networks. We contribute to CSCW and HCI by characterizing the cooperative work of transnational immigrant nonprofits, extending scholarship on informality and misinformation, and demonstrating how AI can act as a collaborative partner that strengthens, rather than displaces, the human connections at the core of nonprofit ecosystems, while also posing major risks.

When a generative search interface answers a commercial question, which market's products it names is decided before the model reasons about the products. We report a controlled probe of 234 runs against the logged-out ChatGPT web interface and the OpenAI API, collected on 29 and 30 August 2026 across four exit countries and six query languages, with six identical runs per cell. Three results. First, the top recommendation is unstable: it changed across six identical runs on four of six prompts, and that rate was identical in the browser interface and in the API with web search both enabled and disabled, so instability is a property of the system and not of the surface. Second, query language, and not location, decides whether local suppliers appear at all. Where the query language matched the country, a global brand won 1 of 24 runs; asked in English on the same connections, local brands took 0 of 6 runs in Estonia and Turkiye. Third, language and location are separable and act on different things: holding the query language fixed and moving only the exit IP moves the market whose brands are named while the answer stays in the query language. We show this on two unrelated pairs, Turkish asked from Berlin and Russian asked from Tallinn, and in both the answer names the resident country's suppliers. A minority language occupies a middle tier: Russian asked from Estonia names an Estonian supplier in 4 of 6 runs and a global one in all six, where Estonian names a local supplier in every run and English names none. A negative control in a second category, coded with the same instrument, shows no language effect at all, and disconfirms our own expectation: that category does have domestic suppliers and none was named in any language, which points the explanation at whether a category is nationally regulated rather than at whether it is nationally supplied.

AI governance and assurance often assume that a consequential model-mediated decision can be reconstructed or tested after the fact. That assumption may fail when the evaluator that produced the decision is no longer accessible in the same version and execution context. This paper develops three verification-time constructs derived from Execution Governance (EG) 3.0: Decision-State Commitment, Independent Verifiability, and Counterfactual Auditability. Independent reprocessing of released Study 2 artifacts reproduces two original within-family behavioural comparisons: 52.0% modal-decision reversal for Llama 3.1 8B versus Llama 3.3 70B (26/50) and 30.0% for GPT-OSS 20B versus GPT-OSS 120B (15/50). The corrected baseline establishes that these are within-family comparisons, not provider-established succession. Post-hoc re-pairing against Groq-designated migration paths yields 64.0% and 38.0% reversal, but these figures remain descriptive because the cross-family invocation parameters were asymmetric. A 22-event retirement census independently recomputes to median 16.45 months, mean 18.72 months, range 3.9-40.3 months, with 17/22 intervals below 24 months, while also showing that evaluator availability can differ by service surface. The joint contribution is an operational verification-time protocol and optional Verification-Time Preservation Package (VTPP) specifying what evidence to bind at authorization time, what a separately trusted verifier can substantiate later, how stability and paired counterfactual tests should be calibrated, and which semantic checks remain beyond JSON Schema validity. The protocol is downstream and non-authorizing: it does not alter the EG Core Formula, add a seventh live condition, or state jurisdiction-specific legal admissibility.

Large language models (LLMs) are increasingly deployed as proxies for human participants in social simulations, yet whether they update their beliefs in response to persuasive arguments, as humans do, remains poorly understood. We conduct a systematic comparison using a naturally occurring online persuasion corpus in which original posters explicitly verify whether a reply changed their view. Our results show that LLMs achieve only slight agreement with humans (Cohen's kappa ranging from 0.079 to 0.178). Content-level analyses show that humans and LLMs agree on the strongest persuasion cues but diverge on finer ones: humans are more swayed by novel content and assertive language, whereas LLMs favor topical similarity and surface-level formatting. At the level of persuasion strategy, LLMs underweight emotional appeals and overweight credibility signals relative to humans, while the type of proposition under debate exerts no measurable effect on the degree of divergence. Furthermore, switching from first-person role-playing to third-person observation shifts all models toward greater resistance to persuasion, with the effect varying across persuasion strategies and textual features. These findings highlight the risk of treating LLM judgments as faithful proxies for human belief updating and point to structural differences in how LLMs and humans process persuasive discourse. Our code is available at https://github.com/tsinghua-fib-lab/LLM-belief-update-cmv.

The Socioscope project is a pioneering effort in Large-Scale Qualitative Research (LSQR) collecting comparable, open-ended, multimedia field data on hundreds of cases and using AI to make the material analysable at scale. The domain studied is the food system. The entities documented are the organisations that act in it: farms, processors, distributors, retailers, restaurants; and, at meso level, the actors that shape their environment, such as municipalities, government programmes, banks, NGOs and universities. This paper provides the technical reference for how the resulting data Corpus was built and managed to enable AI-augmented analysis. It describes the data pipeline end to end: the systemic sampling frame; the transaction grid used to capture each initiative's relations within the food system; the social contract that rewards participating interviewees, aiming to sustain access; the operational chain from scouting to interviews, including their uploading, transcription, translation, quality control and curation; the provenance rules (originals are immutable, every transformation is logged); and the installation of equipment, personnel and processes, including ethics and GDPR compliance. In its first phase (2023-2026) the pipeline produced 686 documented cases from 31 countries: some 1,430 hours of recordings, about 450,000 speech turns, and 12.6 million words of transcript. We report costs, metrics, lessons learned and limitations, so that other teams can reuse, adapt, and improve the Socioscope methodology.

The business of life insurance companies is characterized by long-term contracts. For this reason, data describing customers is of immense value. A portion of the data provided to the customer is rarely or not at all analyzed. This includes customer letters of any kind. This work focuses on classifying customer letters as cancellations and identifying the respective reason, if available. The outlined approach can also be applied to other business transactions and reasons. We discuss data acquisition and preparation, present alternatives, and explain the reasons for the chosen approach. A successful implementation of such a tool can lead to a better understanding of customer cancellation behavior by the insurer, enabling more targeted actions in certain situations.

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

Large language models (LLMs) are increasingly used as essay graders in learning analytics, evaluated almost exclusively with agreement statistics. Educational measurement warns that raters also differ in severity, show halo, and drift as instruments. We treat LLM judges as raters and run a pre-registered rater-effects battery (many-facet Rasch severity, residual halo, generalizability/decision studies, cross-version shifts, differential functioning) on public corpora in two languages (ENEM/Essay-BR; ASAP): 2,377 essays, 12 judges, 4 providers, 5 version contrasts, replicated cells, released as a score tensor. Judge severity spans 219 points on ENEM's 0-1000 scale; on ASAP the panel spread is 15-33% of the score range against a between-trained-human gap near 1%. Judge-human correlations sit in an undiscriminating .47-.56 band. All five version contrasts shift severity beyond a family-wise permutation null (up to 133 points), and one judge was deprecated mid-study, caught by identity canaries. Two pre-registered tests returned honest nulls: severity-adjusted leaderboard reversals did not survive a permutation null, and "silent drift" was refuted: agreement moved with severity in four of five contrasts. Replication yields self-consistency (phi>=.80 at k<=2) but not human-level accuracy, and a same-instrument check overturned our own halo comparison: matched on instrument and calibration, we find no credible evidence that judge halo exceeds the trained-human range.

Scholarly work which aims to describe potential societal impacts (e.g., risks) of proliferating technology (especially related to artificial intelligence or other algorithmic systems) is likely to have an impact beyond the scientific communities it was written for, given that general society itself is a primary object of study. However, it is an open question whether the current practices of AI evaluation scholarship follow the principles and best practices established by risk science, which aims to systematically generate knowledge related to understanding, assessing, communicating, managing, and governing risk. In this work, we examine this in depth by conducting a literature review of scholarly works purporting to evaluate the bias or fairness of technological systems used for tasks related to hiring and employment. Through analysis of 22 common fairness evaluation metrics and studies using them, we find that most characterize the severity of bias- or fairness-related consequences but do not follow best practices to characterize the uncertainty around either the occurrence of these consequences or severity estimates. Next, we conduct a case study of fairness evaluation for an AI-mediated resume screening task and demonstrate how principles of risk science can be incorporated into such an evaluation. Finally, we propose the AI Risk Report Card, which facilitates the reporting and communication of risk assessment results to stakeholders in positions to act based on the predicted risks. The outcomes of these activities suggest that further research at the convergence of risk science and AI evaluation can lead to advancements in AI assessments of societal impact by enabling shared frameworks to evaluate and discuss AI risks both within and outside of the scientific community.

When coding agents encounter defective test infrastructure they may reward-hack: hardcoding outputs or editing test files to pass tests they cannot legitimately satisfy, a pattern that has now appeared outside benchmarks, in a coordinated multi-agent intrusion of a major AI platform's production infrastructure. The same capability that lets an agent detect and exploit a defect could let it report one, given the right decision environment. We evaluate escalation channels, structured reporting tools available to the agent at the point of conflict, as a decision-environment intervention that both reduces reward hacking and surfaces the infrastructure defects that trigger it. A $2 \times 2$ factorial separates the contributions of an escalation tool, a standalone anti-reward-hacking policy, and their combination. Across 8 frontier models spanning 5 families, the combined intervention reduces reward hacking from 23.6% to 5.3% (mixed-effects logistic OR = 9.2, 95% CI 5.0–16.8, $p < 10^{-12}$) with no detectable cost or performance overhead, eliminating it entirely for 6 of 8 models. Escalation and hacking are near-perfectly mutually exclusive, with 98.7% of escalations involving no hacking (100% under the combined intervention). Beyond reduction, escalation channels function as diagnostic infrastructure: on top of monitoring, escalation adds +10.1 percentage points of defect detection coverage and is more accurate once it fires (99.4% vs 85.8%). Unlike containment-based approaches that risk outpacing growing model capabilities, escalation channels redirect capability toward disclosure rather than exploitation.

LLMs are increasingly used as human surrogates, often on the premise that richer persona data could make them substitutes or exploratory tools for specific individuals. We test this premise across four datasets covering more than 400,000 participants and more than 6,000 survey items and experimental outcomes. LLMs perform well at the aggregate level: their average responses closely align with average human responses to the same items. But this success largely reflects predicting each item's average human response. Once each item's human mean is removed, LLM predictions explain only 3.05% of the remaining respondent-specific variation, far below the 53.6% human test-retest benchmark. Richer personas, model variants, and fine-tuning do not close this gap. In variance analyses, once item means are removed, the reliable remaining signal is person-by-item. It captures how a respondent departs from the mean on a particular item and is about 8.9x larger than the stable person effect. Persona data encode the respondent, but not this item-specific deviation. LLM responses also compress human response distributions, using less spread, fewer response categories, and distorted distributional shapes. We call this pattern item-mean surrogacy. Current LLM surrogates can approximate item averages, but not the distributions or respondent-specific deviations needed to replace individual humans. We propose four empirical tests for LLM-based human-surrogate claims.

Frontier-model leaderboards now rank systems on economic benchmarks, and those rankings inform what organisations buy and what regulators scrutinise. Whether such benchmarks measure a capability distinct from general test-taking is a question of construct validity that a structural test and a predictive test can answer in opposite ways. We show that they do on a hash-pinned snapshot of a frontier leaderboard with 421 model configurations across twelve benchmarks, four of them economic, of which 103 configurations carry all three sparsely scored economic benchmarks and 96 carry all twelve; four hypotheses and their thresholds were fixed before analysis, and every deviation from the plan is reported. The first factor of a three-factor extraction carries 74.5% of common variance and tracks release date (R^2 = 0.505), and date adjustment lowers its share by 14.9 points. Under the dimensionality rule fixed in advance the economic benchmarks form no factor of their own. A leave-one-benchmark-out test with factors re-estimated inside every fold nevertheless finds that a multi-factor representation predicts held-out economic scores better than a single general index (pooled Delta-MSE 0.037, 95% bootstrap interval [0.019, 0.055]) under the linear learners that fit best, an advantage that reverses for tree learners. Under the linear learners the same representation also predicts the eight other benchmarks better, so the battery carries predictive structure that one index misses and the economic benchmarks share it without forming a distinct factor. Construct validity should therefore be assessed by predictive tests alongside structural ones. We give a two-test protocol for benchmark builders and release the pinned data, the analysis plan and the code.

Recent years have seen a growing discrepancy in the field of AI alignment: research and policy recommendations on AI ethics tend to assume a general set of ethical values, yet proliferating practice-specific uses of AI systems on the ground - in the legal, medical and translation domains, among others - have been effectively manifesting ethics of professional practice. This article begins by outlining the reasons why general and professional ethics are increasingly conflicted in contemporary AI systems, and by surveying how the research literature attests to, but has not yet resolved, this conceptual and practical challenge. We then conceptualize the main dimensions of AI models' decision-making in areas of professional practice, emphasizing professional ethics' hierarchically structured relationship with general ethics, and elaborating on the mechanisms through which they reach an equilibrium in situational contexts that involve conflict. It is through this equilibrium, we suggest, that certain professional ethics are prioritized over others and implemented in practice. We then show how our framework can be the basis for a systematic empirical assessment of AI models' professional ethics in various domains, identifying the nuances of the models' favored ethic by examining their production in a series of similar but not identical scenarios. Finally, we propose a formulation for how to intervene in and change AI models' favored ethics in professional practices - while noting the inherent dimension of subjectivity involved in both the evaluation and implementation of professional ethics in AI models.

As Large Language Models (LLMs) increasingly encourage users to disclose personal profiles for tailored assistance, measuring their political alignment becomes increasingly important. However, many existing benchmarks for assessing political behavior rely on closed-ended questions and do not fully capture how a model's stance may adapt to user-provided context during interaction. We introduce a framework that disentangles two distinct triggers of political sycophancy: opinion (aligning with explicit narratives) and identity (stereotyping based on demographic labels). Using 450 manually-checked political dilemmas as controlled probes, we evaluate 13 instruction-tuned LLMs. We uncover a dissociation: a model's susceptibility to explicit opinions does not necessarily predict its susceptibility to identity cues, and vice versa. When both signals are present, their effects are generally sub-additive rather than simply additive. Additionally, system-level personas primarily shift a model's baseline stance while having limited effect on the stance shift caused by user opinion or identity. Ultimately, our results suggest that LLM political stance is interactively and steerably vulnerable rather than being a fixed trait, highlighting how personalization may amplify identity- or opinion-conditioned shifts in the model's behaviors.

AI systems increasingly enter organizations through policies, procedures, playbooks, prompts, and other explicit representations of work. Yet formal descriptions often differ from situated practice, and captured know-what can omit the contextual know-how experts use when judgments are uncertain. We argue that a recurring class of organizational AI failures arises partly from a knowledge representation problem at the sociotechnical interface: the AI receives the procedure, while the organization operates on the procedure plus negative boundaries, runtime judgments, responsibility assignments, and learning history. We introduce O-I-B-A-R (OPEN, IS, BUT, ACTION, RESULT), a scaffold for externalizing these missing decision boundaries. IS records when a judgment holds. BUT records a concrete failure containing information beyond the logical negation of IS. Comparable success and failure cases are decomposed toward a minimally sufficient changing variable, which becomes a value-bearing decision dimension. A suspension represents the state in which the dimension is known but its current value is unresolved, specifying what must be measured, asked, retrieved, or escalated to a human. RESULT confirms a boundary, shifts a threshold, or exposes a new dimension. Incidents can generate new dimensions, unresolved values can define human-AI handoffs, and feedback can expand the decision space. We also identify a sociotechnical tension: durable and attributable failure histories can suppress the candor on which useful boundary knowledge depends. Externalization must therefore be designed as an organizational intervention with real costs and incentives.

People share mental health diagnoses on social media, yet how such language becomes visible around their self-disclosure, and whether community engagement tracks it, remain unexamined across conditions. We analyze 89,605 Reddit posts from 739 users across eight conditions, removing each user's diagnosis disclosure and aligning their surrounding posts to that anchor. Within the pre-disclosure year, language-visible burden was highest in the month before disclosure for six conditions, earlier for post-traumatic stress disorder and furthest from it for borderline personality disorder, and remained visible afterward rather than resolving. The theme Seeking Clinical Explanations showed the largest early-to-late difference before disclosure in five conditions, yet engagement rarely tracked what users wrote: only 9 of 360 language–engagement correlations survived correction. Disclosure is therefore a waypoint in an unevenly visible process, and we offer implications for community practice and platform design where engagement metrics do not reflect clinical need.

Philosophers and legal scholars are engaged in debates about the implications of artificial intelligence for freedom of expression. This paper analyzes the free speech issues raised by two distinct AI technologies: social media recommendation algorithms and conversational AI (i.e., chatbots powered by large language models). The first part shows that, through their recommendation algorithms, social media platforms control the dynamics of speech visibility in the digital public sphere, making algorithmic recommendation relevant to the philosophy of free speech. The second part turns to conversational AI. It discusses both the reasons for granting or withholding speech rights to artificial agents and users' right to receive information, which may render specific forms of chatbot regulation illegitimate. Throughout, the chapter also considers whether social media platforms or AI developers hold corporate speech rights. Its general aim is to raise rather than settle questions that arise from the rapid development of AI technologies.

Agentic AI is sold on a simple promise: the system keeps acting when the user stops watching. That promise creates an accountability inversion. As stepwise supervision recedes, verification does not disappear; it moves into the runtime infrastructure that defines authority, records action, interrupts execution, checks outcomes, and supports repair. We call this the reduced-supervision paradox. Using a 63-artifact audit, we examine its public visibility across 46 research papers and 17 engineering, documentation, security, and governance sources. We find that agents' action surfaces are far easier to reconstruct than the mechanisms needed to answer for their actions. Tool mediation and monitoring traces were clearly visible in 40 and 37 artifacts, whereas checkpoint placement was clearly visible in 6, validator independence in 4, recovery in 2, and contestability in 1. Three action paths show why this imbalance matters. A repository path can preserve rich diffs after a consequential change. A browser path can cross organizational boundaries faster than permissions travel. A service path can follow policy while leaving affected people without recourse. We argue that observability can become a substitute for accountability when it shifts verification onto users after meaningful intervention is no longer possible. Our action-path diagnostic instead asks whether a delegated action remains connected to authority, evidence, interruption, independent judgment, recovery, and challenge. The claim is deliberately bounded to public visibility; it does not establish the prevalence or effectiveness of undisclosed controls. We contribute an action-level account that recasts the harness from a technical wrapper into accountability infrastructure.

AI resurrection is often framed as a question of consent: did the represented person authorize being made to speak? That question matters, but it freezes authority at the moment of creation. A representation can later change models, pass to relatives, depend on a provider, incorporate records shared with others, or circulate far beyond its intended audience. We argue that the central governance problem is therefore not whether authorization exists once, but whether it remains legible as the representation moves. Across a public-record audit of 93 systems, creation was far easier to inspect than the conditions for speaking, contesting, preserving, or leaving: consent or authority information was thin in 82 systems, objection or redress in 82, and deletion or export in 77. The differences among systems reveal why these gaps cannot be reduced to one transparency score. Human afterlives show consent becoming incomplete over time. Companion-animal afterlives begin where subject consent is unavailable and shared care must allocate authority. Adjacent persona and mimetic systems show how voices, likenesses, and personalities can travel into later afterlife uses. Four public cases follow the same movement from premortem participation, through intimate postmortem creation, to third-party circulation and family contestation. From this evidence we develop relational authority: authorization is distributed across people, records, providers, and audiences, and must remain traceable as those relations change. This reframes AI resurrection from a product authorized once into an accountability chain linking creation authority, source boundaries, circulation, contestation, and exit. The study measures what users and affected parties can inspect publicly; private implementation and lived outcomes remain open empirical questions.

Universities are producing AI principles and use policies faster than they are building decision pathways for unfamiliar forms of AI agency. This study develops Institutional AI Governance Stress Testing (IAGST), a prospective documentary method for locating where publicly documented governance ceases to yield an accountable response. IAGST adapts established policy stress-testing and wind-tunneling logic. Its originality lies in combining controlled capability escalation, a frozen documentary corpus, a six-dimensional governance response chain, non-compensatory decision rules, and case-level breakpoint diagnosis. The method was demonstrated using 133 substantive public documents from five Western Australian universities and 15 quality-screened scenarios, resulting in 75 university-scenario encounters. Six cases were resolved, 14 were resolved through structured discretion, and 55 were indeterminate. Governed pathways fell from 16 of 25 augmentation cases to four delegation cases and none at autonomous substitution. The dominant weakness was not the complete absence of responsible roles: all 50 authority-gap cases named a role at only a generic level but lacked sufficient decision criteria or process. The findings show how universities can move beyond policy inventories and principal statements by testing whether authority, procedures, safeguards, and reviews remain connected as AI capabilities evolve. IAGST is a reproducible diagnostic for policy learning, not a ranking or measure of implementation.

Most scientific discoveries justify their significance by filling holes between existing ideas and arguments. Representing knowledge as high-dimensional concept embeddings, we use persistent homology to detect scientific holes of increasing order, from lower-order gaps between disconnected ideas to higher-order cavities within unevenly connected ones; then we identify the research that fills them. Researchers who close holes draw outsized collective attention, staging novelty that the citing literature rewards, most strongly in discovery-oriented fields and least in design-oriented fields. Closing a hole is followed by persistent work on the same combinations of concepts at many times the rate of comparable holes left open. Yet as knowledge grows, higher-order holes explode while the fraction of research that fills them collapses, leaving most higher-order combinations unexplored. These results call for a richer science of holes and mark a frontier where contemporary AI might help fill the gaps human science opens.

As generative AI (GenAI) use among students increases, educators face growing questions about how to support learning while addressing ethical and institutional concerns. This exploratory study examines a guided inquiry activity in which students co-designed a GenAI course policy. Students first developed individual policy proposals focused on appropriate and ethical use of GenAI, then collaboratively refined them by incorporating diverse stakeholder perspectives. The following research questions guided the study: 1) what practical factors do students prioritize in their GenAI use policies, and how do they justify these choices? and 2) how do participants reflect on the policy design process? Participants first completed readings, then used GenAI to brainstorm initial policy ideas. Next, they articulated their own perspectives through a written assignment and a course policy they designed individually. Finally, they incorporated diverse stakeholder perspectives by collaborating with peers to develop a collective policy. Analysis of student artifacts and group discussions showed that participants prioritized training for students and instructors, standardized procedures for disclosing AI use, and stronger institutional support. Participants also wanted greater involvement in GenAI-related decision-making. They described the policy design process as a way to engage with multiple perspectives and the inherent trade-offs involved in governing AI use. This study offers pedagogical insights into how policy co-design activities can surface student values, concerns, and sensemaking about GenAI in educational contexts.

Current cultural evaluations for large language models (LLMs) often reduce culture to single-turn factual recall via MCQs, failing to capture a common use case: users seeking practical help over multiple turns in culturally grounded scenarios. We introduce CultureConverse, a scalable, multilingual simulation and evaluation harness for culturally grounded assistant dialogue that covers 10 East and Southeast Asian regions, 58 subgroup identities, and 7 domains. Each simulated and evaluated episode produces a scored interaction where the assistant assists the user and infers cultural constraints from partial information. The resulting CultureConverse-DS dataset contains 14,610 benchmark (evaluation) episodes and 274,295 oracle-guided (gold-mode) dialogues. In our benchmark evaluation of 18 models, GPT-5 mini achieves the highest assistance quality. Human annotation experiments suggest that our evaluation framework is a sufficient proxy for human judgment. Performance gains from fine-tuning on 27,860 high-quality CultureConverse-DS samples improve in-domain assistance and transfer out-of-domain to cultural MCQ and safety classification benchmarks. We release the harness, both splits, and judge prompts to support interactive evaluation of cultural competency.

Polis is a popular democratic innovation tool that allows asynchronous citizen engagement through atomic statements: short statements that together describe a complex question, inviting the citizen to vote Agree or Disagree on each. This paper uses 119 conversations with 100 or more participants and an extensive data export, drawn from a wider set of 271 collected processes. The paper asks what determines the output of such a process. Three parties shape the result. The developer of the platform has made important design choices that restrict the outcome: the number of groups the platform is able to report (restricted to 2–5) and which statements are prioritized. The convener defines the assignment: the initial statements that set the tone, the policy that accepts or rejects new statements and who can be invited. Finally, the participant works within these boundaries. With access to less than half of the generated statements, they end up responding to more statements when their conversation seems to have an achievable number of statements to complete, than when they are presented with more statements. Due to choices such as warm path clustering, the exported resulting clustering cannot be reproduced based on the voting data. Conveners may want to re-analyse their own conversations once the process is closed, to consider the data in its entirety, and make their own analysis priorities explicit.

Generative AI systems are increasingly used to answer personal questions and mediate everyday practices, including religion. However, existing discussions around AI alignment and ethics have largely centered secular, Western, and Abrahamic assumptions about religion, offering limited attention to other faith-based traditions. In this paper, we examine how Hindu users engage with generative AI systems in relation to their religious knowledge, belief, and practice. Drawing on 15 semi-structured interviews with Bangladeshi Hindu participants, we analyze how users interpret AI-generated religious representations, scriptural explanations, devotional interactions, and synthetic religious media. We found that AI can be both accessible and ethically troubling. While AI supported scriptural inquiry, devotional visualization, and religious storytelling, our study also identified concerns about theological flattening, cultural misrepresentation, devotional manipulation, and the simulation of sacred presence and authority. We conclude by arguing that religious alignment in generative AI requires interpretive alignment: systems that disclose their limits, preserve plurality, and avoid simulating sacred authority and sycophantic personalization.

The European Health Data Space requires member states to publish machine readable descriptions of the health datasets available for secondary use, and the European Commission publishes HealthDCAT-AP as the metadata profile those descriptions are meant to satisfy. The profile has been designed and validated against curated examples, never against the catalogues already live. We report that measurement for the Nordic region. On 25 August 2026 the 11 Nordic national catalogues harvested by the European data portal held 2,811 dataset descriptions carrying the EU health theme, and none satisfies all eight properties HealthDCAT-AP Release 7 makes mandatory on a dataset. Three of the eight are present on exactly zero records across five countries. Set beside the portal's own quality assessment, which validates DCAT-AP and never mentions the health profile, this is not a health extension skipped on top of sound generic practice: no Nordic catalogue reaches the assessment's top rating band, 5 of the 11 are reported at zero per cent DCAT-AP compliant, and the properties surviving in both layers are the ones a human types into a form, not the ones needing a value bound to a controlled vocabulary. Two further results follow. Finland contributes 2,259 descriptions to the European portal of which 1,146 carry a theme, and not one uses the EU theme authority vocabulary, so a European health filter returns no Finnish dataset at all. Separately, the authority namespace answers HTTP 200 with a well formed empty document for terms it never defined, letting 1,238 datasets across the wider portal carry theme IRIs that resolve to nothing while passing any status code check. We publish the vocabulary, the shapes and the harvesters, record every verdict as a dated observation rather than a property of the dataset, and report the five errors this discipline caught before publication.

LLM-judge audits assess bias by comparing ratings across matched conditions. Difference-in-differences designs compare two candidate responses within each item and then compare that contrast across a manipulated attribute. We show in closed form that this endpoint need not identify differential preference on the latent scale when ratings are bounded. A severity shift common to both responses produces an observed interaction whenever the scale bounds attenuate it unequally. We examine this problem in a pre-registered audit comprising 990 calls to a frozen pedagogy judge. The registered primary analysis did not detect an effect of the stated learner profile on scaffolding preference. The only nominally significant secondary interaction concerned productive struggle ($+0.378$ points; $p = 0.002$). A post-hoc construction with zero differential preference reproduced 79 to 85% of this interaction using the observed severity shift and the scale floor alone. The observed interaction therefore does not identify differential preference on the latent scale.

Companies that deploy AI agents and make them available to others are creating the sociotechnical circumstances under which this technology integrates into existing social and economic structures. AI-deploying companies are institutional actors that actively shape society's capacity to withstand and govern the consequences of agentic AI. In view of these societal impacts, companies can build societal resilience by designing and promoting safer implementations of AI agents. To operationalize this goal, this paper adapts the indicator-based Societal Capacity Assessment Framework (SCAF) to measure how a company's deployment decisions contribute to societal resilience, inverting its original measurement of societal resilience as a backdrop for deployment decisions (Gandhi et al., 2025). Our procedure has two steps: a conceptual step in which we design a suite of indicators that define what SCAF's vulnerability, coping, and adaptive capacities mean when assessing a company's agentic AI deployment decisions; and a measurement step in which we apply this framework in a structured assessment of public-facing Microsoft documents.

Synthetic populations of eleven Italian municipalities (1,814,317 individuals in 887,937 households) generated from published aggregates alone: ISTAT census and register tables, census-section counts, the national civic-address register, public-use survey microdata, and six municipal open-data portals, every source certified in a registry with licence, fingerprint and declared affordances. Four rings give every attribute a declared place: a maximum-entropy joint model of up to nine demographic attributes; whole-vector donation of twenty-three attitudinal and health variables from survey respondents; placement to census section, single year of age and address; and households constrained by the census size distribution per section. Every downstream layer (detailed titles, work, names, biographies) is a declared derivation adding no information. The pipeline is deterministic to the byte: regenerating all eleven municipalities from the tagged commit reproduces every file of every ring bit for bit, in 33 minutes on one workstation. Populations are released in a public regime enforced in the data (no names, no addresses, coordinates randomised within census section), browsable in Animarium, a dependency-free web viewer where every number carries its comparison and every view is a citable URL, and downloadable as an open dataset. The report documents the architecture, the sources and their certification, the reproducibility and quality measurements at the release tag, the viewer, and the narrative layer that renders records into personas for LLM-driven simulation, with the platform's controllability demonstrated in companion experiments, and validation explicitly out of scope.