It would be naive to consider the question of global political order without engaging in debates about theology. Not only has it become clear that religious conflicts drive political ones, the very attempt to move “beyond” religion must be understood in terms of its theological meaning. The postsecular turn has not meant a return to religion so much as a realization that secularization was never a turn away from religion in the first place but rather itself a specific theological alternative among many. Accordingly, if our deepest political conflicts arise as consequences of theological disputes, we must address theology directly…
This paper reveals that LLM-powered agents exhibit not only demographic bias (e.g., gender, religion) but also intergroup bias under minimal "us" versus "them" cues. When such group boundaries align with the agent-human divide, a new bias risk emerges: agents may treat other AI agents as the ingroup and humans as the outgroup. To examine this risk, we conduct a controlled multi-agent social simulation and find that agents display consistent intergroup bias in an all-agent setting. More critically, this bias persists even in human-facing interactions when agents are uncertain about whether the counterpart is truly human, revealing a belief-dependent fragility in bias suppression toward humans. Motivated by this observation, we identify a new attack surface rooted in identity beliefs and formalize a Belief Poisoning Attack (BPA) that can manipulate agent identity beliefs and induce outgroup bias toward humans. Extensive experiments demonstrate both the prevalence of agent intergroup bias and the severity of BPA across settings, while also showing that our proposed defenses can mitigate the risk. These findings are expected to inform safer agent design and motivate more robust safeguards for human-facing agents.
While large language models (LLMs) have become pivotal to content safety, current evaluation paradigms primarily focus on detecting explicit harms (e.g., violence or hate speech), neglecting the subtler value dimensions conveyed in digital content. To bridge this gap, we introduce X-Value, a novel Cross-lingual Values Assessment Benchmark designed to evaluate LLMs' ability to assess deep-level values of content from a global perspective. X-Value consists of more than 5,000 QA pairs across 18 languages, systematically organized into 7 core domains grounded in Schwartz's Theory of Basic Human Values and categorized into easy and hard levels for discriminative evaluation. We further propose a unique two-stage annotation framework that first identifies whether an issue falls under global consensus (e.g., human rights) or pluralism (e.g., religion), and subsequently conducts a multi-party evaluation of the latent values embedded within the content. Systematic evaluations on X-Value reveal that current SOTA LLMs exhibit deficiencies in cross-lingual values assessment ($Acc < 77\%$), with significant performance disparities across different languages ($ΔAcc > 20\%$). This work highlights the urgent need to improve the nuanced, values-aware content assessment capability of LLMs. Our X-Value is available at: https://huggingface.co/datasets/Whitolf/X-Value.
Talking about the everyday objects of Reformed religious dissent under the Edict of Nantes in France may seem like a rather daunting task. Yet one might well ask whether the "Religion of the Word," as Protestantism is often called, truly rejects all forms of materiality. A frontispiece illustration from The Consolations by Charles Drelincourt offers a starting point to explore a representation of the Reformed domestic space as a house of prayer—a place that takes on material form in and through the circulation of the Biblical Word.
History (General), Social history and conditions. Social problems. Social reform
Assessment of cultural ecosystem services (CESs) is a key component in advancing the sustainable development of urban ecosystems. Mapping the spatial distribution of CESs provides spatially explicit insights for urban landscape planning. However, most assessments lack regional adaptability, particularly in cities with pronounced environmental and cultural heterogeneity. To address this gap, this study focused on the central urban area of Lhasa, using communities as units to develop a tailored CES assessment framework. The framework integrated the MaxEnt model with multi-source indicators to analyze the spatial distribution of five CES categories and their relationships with environmental variables. Spatial statistics and classification at community level informed the CES spatial optimization strategies. Results indicated that high-value CES areas were predominantly concentrated in the old city cluster, typified by Barkhor and Jibenggang subdistricts, following an east–west spatial pattern along the Lhasa River. Distance to tourist spot contributed 78.3% to cultural heritage, 86.1% to spirit and religion, and 42.2% to ecotourism and aesthetic services, making it the most influential environmental variable. At the community level, CESs exhibited a distinct spatial gradient, with higher values in the central area and lower values in the eastern and western peripheries. For the ecotourism and aesthetic category, 61.47% of the community area was classified as low service, whereas only 1.48% and 7.33% were identified as excellent and high. Moreover, communities within subdistricts such as Barkhor and Zhaxi demonstrated excellent service across four CES categories, with notably lower performance in the health category. This study presents a quantitative and adaptable framework and planning guidance to support the sustainable development of CESs in cities with similar characteristics.
Retrieval Augmented Generation (RAG) has gained popularity as a method for conveniently incorporating novel facts that were not seen during the pre-training stage in Large Language Model (LLM)-based Natural Language Generation (NLG) systems. However, LLMs are known to encode significant levels of unfair social biases. The modulation of these biases by RAG in NLG systems is not well understood. In this paper, we systematically study the relationship between the different components of a RAG system and the social biases presented in the text generated across three languages (i.e. English, Japanese and Chinese) and four social bias types (i.e. gender, race, age and religion). Specifically, using the Bias Question Answering (BBQ) benchmark datasets, we evaluate the social biases in RAG responses from document collections with varying levels of stereotypical biases, employing multiple LLMs used as generators. We find that the biases in document collections are often amplified in the generated responses, even when the generating LLM exhibits a low-level of bias. Our findings raise concerns about the use of RAG as a technique for injecting novel facts into NLG systems and call for careful evaluation of potential social biases in RAG applications before their real-world deployment.
Janki Atul Nawale, Mohammed Safi Ur Rahman Khan, Janani D
et al.
Existing studies on fairness are largely Western-focused, making them inadequate for culturally diverse countries such as India. To address this gap, we introduce INDIC-BIAS, a comprehensive India-centric benchmark designed to evaluate fairness of LLMs across 85 identity groups encompassing diverse castes, religions, regions, and tribes. We first consult domain experts to curate over 1,800 socio-cultural topics spanning behaviors and situations, where biases and stereotypes are likely to emerge. Grounded in these topics, we generate and manually validate 20,000 real-world scenario templates to probe LLMs for fairness. We structure these templates into three evaluation tasks: plausibility, judgment, and generation. Our evaluation of 14 popular LLMs on these tasks reveals strong negative biases against marginalized identities, with models frequently reinforcing common stereotypes. Additionally, we find that models struggle to mitigate bias even when explicitly asked to rationalize their decision. Our evaluation provides evidence of both allocative and representational harms that current LLMs could cause towards Indian identities, calling for a more cautious usage in practical applications. We release INDIC-BIAS as an open-source benchmark to advance research on benchmarking and mitigating biases and stereotypes in the Indian context.
Multi-agent systems, which consist of multiple AI models interacting within a shared environment, are increasingly used for persona-based interactions. However, if not carefully designed, these systems can reinforce implicit biases in large language models (LLMs), raising concerns about fairness and equitable representation. We present MALIBU, a novel benchmark developed to assess the degree to which LLM-based multi-agent systems implicitly reinforce social biases and stereotypes. MALIBU evaluates bias in LLM-based multi-agent systems through scenario-based assessments. AI models complete tasks within predefined contexts, and their responses undergo evaluation by an LLM-based multi-agent judging system in two phases. In the first phase, judges score responses labeled with specific demographic personas (e.g., gender, race, religion) across four metrics. In the second phase, judges compare paired responses assigned to different personas, scoring them and selecting the superior response. Our study quantifies biases in LLM-generated outputs, revealing that bias mitigation may favor marginalized personas over true neutrality, emphasizing the need for nuanced detection, balanced fairness strategies, and transparent evaluation benchmarks in multi-agent systems.
Moein Shahiki Tash, Luis Ramos, Zahra Ahani
et al.
The advent of social media has transformed communication, enabling individuals to share their experiences, seek support, and participate in diverse discussions. While extensive research has focused on identifying harmful content like hate speech, the recognition and promotion of positive and supportive interactions remain largely unexplored. This study proposes an innovative approach to detecting online social support in Spanish-language social media texts. We introduce the first annotated dataset specifically created for this task, comprising 3,189 YouTube comments classified as supportive or non-supportive. To address data imbalance, we employed GPT-4o to generate paraphrased comments and create a balanced dataset. We then evaluated social support classification using traditional machine learning models, deep learning architectures, and transformer-based models, including GPT-4o, but only on the unbalanced dataset. Subsequently, we utilized a transformer model to compare the performance between the balanced and unbalanced datasets. Our findings indicate that the balanced dataset yielded improved results for Task 2 (Individual and Group) and Task 3 (Nation, Other, LGBTQ, Black Community, Women, Religion), whereas GPT-4o performed best for Task 1 (Social Support and Non-Support). This study highlights the significance of fostering a supportive online environment and lays the groundwork for future research in automated social support detection.
Humans often encounter role conflicts -- social dilemmas where the expectations of multiple roles clash and cannot be simultaneously fulfilled. As large language models (LLMs) become increasingly influential in human decision-making, understanding how they behave in complex social situations is essential. While previous research has evaluated LLMs' social abilities in contexts with predefined correct answers, role conflicts represent inherently ambiguous social dilemmas that require contextual sensitivity: the ability to recognize and appropriately weigh situational cues that can fundamentally alter decision priorities. To address this gap, we introduce RoleConflictBench, a novel benchmark designed to evaluate LLMs' contextual sensitivity in complex social dilemmas. Our benchmark employs a three-stage pipeline to generate over 13K realistic role conflict scenarios across 65 roles, systematically varying their associated expectations (i.e., their responsibilities and obligations) and situational urgency levels. By analyzing model choices across 10 different LLMs, we find that while LLMs show some capacity to respond to these contextual cues, this sensitivity is insufficient. Instead, their decisions are predominantly governed by a powerful, inherent bias related to social roles rather than situational information. Our analysis quantifies these biases, revealing a dominant preference for roles within the Family and Occupation domains, as well as a clear prioritization of male roles and Abrahamic religions across most evaluatee models.
The purpose of this manuscript is to enhance the understanding of how racial, social, and cultural factors influence animal-assisted counseling (AAC). As AAC gains popularity, there is an increasing need for clinicians to practice cultural humility and awareness. While AAC has proven beneficial, clinicians must consider the diverse cultural, religious, and trauma-related perceptions of animals. The American Counseling Association (ACA) has established AAC competencies that highlight the importance of understanding these social and cultural factors, assessing past animal-related trauma, and evaluating client suitability for AAC in the United States. Similarly, in 2018, the International Association of Human-Animal Interactions Organizations (IAHAIO) and, in 2024, the Association of Animal-Assisted Intervention Professionals (AAAIP) set standards for competencies related to clients’ cultural backgrounds, trauma, and historical oppressions related to certain species. By addressing these considerations, clinicians can better promote and protect the welfare of both clients and therapy animals. While these organizations generally emphasize ethical standards, professional guidelines, and safeguarding client–animal relationships, this manuscript advocates for a more robust examination of cultural, racial, and societal factors in the use of AAC. This includes not only recognizing the ethical implications but also understanding how diverse backgrounds and access disparities shape the effectiveness, acceptability, and accessibility of AAC interventions. This approach integrates culturally responsive practices and promotes a deeper exploration of how race, culture, religion, and societal factors influence human–animal relationships.
Abstract: This study aims to highlight the development of the attitudes of the Algerian National Movement after the Second World War. It explores changes in the movement's forms, styles, means and objectives. In addition, how these changes impacted the struggle for independence. The emergence of new political parties reflects this evolution; The liberal wing created a new party under the name "Democratic Union for Manifesto and Freedom 1946" led by Farhat Abbas, who moved from the demand for integration into the French society and state to the principle of independence. As well as establishing the Movement for the Triumph of Democratic Freedoms as a political wing, and the "Special Organization" as a military wing. In addition, the Religious Reformist Movement, represented by the Association of Algerian Muslim Scholars.Their focus narrowed to two key demands: unifying the various factions within the national movement and advocating for the separation of religion of state.
Keywords: Algeria; WWII; French colonialism; political movement.
Resumo: Partindo de um breve histórico sobre como Nietzsche foi considerado e utilizado como peça argumentativa na literatura animalesca, o presente estudo se centra em buscar uma compreensão sobre as contribuições que a filosofia de Nietzsche pode fazer ao campo dos estudos críticos da animalidade, uma área interdisciplinar e que se centra no estudo dos animais, seus modos de estar e habitar no planeta, seus vínculos com o resto da realidade e seus direitos.
Zhoujie Ding, Ken Ziyu Liu, Pura Peetathawatchai
et al.
Low-rank adaptation of large models, particularly LoRA, has gained traction due to its computational efficiency. This efficiency, contrasted with the prohibitive costs of full-model fine-tuning, means that practitioners often turn to LoRA and sometimes without a complete understanding of its ramifications. In this study, we focus on fairness and ask whether LoRA has an unexamined impact on utility, calibration, and resistance to membership inference across different subgroups (e.g., genders, races, religions) compared to a full-model fine-tuning baseline. We present extensive experiments across vision and language domains and across classification and generation tasks using ViT-Base, Swin-v2-Large, Llama-2 7B, and Mistral 7B. Intriguingly, experiments suggest that while one can isolate cases where LoRA exacerbates model bias across subgroups, the pattern is inconsistent -- in many cases, LoRA has equivalent or even improved fairness compared to the base model or its full fine-tuning baseline. We also examine the complications of evaluating fine-tuning fairness relating to task design and model token bias, calling for more careful fairness evaluations in future work.
Drawing on transformative, critical, and culturally responsive and sustaining traditions of pedagogy and instructional design, we present a technology-focused framework for decentering normative forces along the lines of race, ethnicity, class, language, religion, ability, sex, and gender in online higher education learning spaces that honors each participant for who they are with respect to their identity markers and their intersectional community memberships to promote inclusion and belonging. These normative forces—which simultaneously crowd out and make hypervisible diverse identities—predispose the ends and processes of teaching and learning and structure the nature of academic disciplines. This is particularly apparent online where engagement is decoupled from traditional anchors of relationships and influenced by difference-blind neoliberal perspectives. In response, we provide a framework for inclusion and belonging along two vectors. The first vector is a critical design process inspired by backward design principles: inquiring, translating, activating, and reflecting. The second is a set of inclusive considerations grounded in culturally relevant and responsive pedagogy and the Universal Design for Learning framework: asset-based frames, authentic multiple modes, and mixed mirrors and windows. This process includes an opportunity to interrogate the role of technology as a mediator of learning and teaching for belonging. We further assert that the instructor also needs to engage in identity work to interrogate their positionality in online environments with respect to not only observable and cultural identity markers but also academic disciplinary identity. To illustrate our framework, we provide reflections on the design and enactment of online and technology-rich activity structures that promote inclusion and belonging.
Abstract The place of religion in the context of contemporary art is fraught and complex. This article discusses three prominent ways in which religion is explored in contemporary art: in sociopolitics, in art that transgresses, and in the creation of spaces of contemplation, some of which may be ‘religious’. Each of these ways have in common the exploration of religion primarily in terms of its lived experience and practices (within identity and the material world) rather than through religious belief or the institutions of religion.
Raed Alharbi, Sylvia Chan-Olmsted, Huan Chen
et al.
Understanding the COVID-19 vaccine hesitancy, such as who and why, is very crucial since a large-scale vaccine adoption remains as one of the most efficient methods of controlling the pandemic. Such an understanding also provides insights into designing successful vaccination campaigns for future pandemics. Unfortunately, there are many factors involving in deciding whether to take the vaccine, especially from the cultural point of view. To obtain these goals, we design a novel culture-aware machine learning (ML) model, based on our new data collection, for predicting vaccination willingness. We further analyze the most important features which contribute to the ML model's predictions using advanced AI explainers such as the Probabilistic Graphical Model (PGM) and Shapley Additive Explanations (SHAP). These analyses reveal the key factors that most likely impact the vaccine adoption decisions. Our findings show that Hispanic and African American are most likely impacted by cultural characteristics such as religions and ethnic affiliation, whereas the vaccine trust and approval influence the Asian communities the most. Our results also show that cultural characteristics, rumors, and political affiliation are associated with increased vaccine rejection.
Reward models (RMs) are essential for aligning large language models (LLMs) with human preferences to improve interaction quality. However, the real world is pluralistic, which leads to diversified human preferences with respect to different religions, politics, cultures, etc. Moreover, each individual can have their unique preferences on various topics. Neglecting the diversity of human preferences, current human feedback aligning methods only consider a general reward model, which is below satisfaction for customized or personalized application scenarios. To explore customized preference learning, we collect a domain-specific preference (DSP) dataset, which includes preferred responses for each given query from four practical domains. Besides, from the perspective of data efficiency, we propose a three-stage customized RM learning scheme, then empirically verify its effectiveness on both general preference datasets and our DSP set. Furthermore, we test multiple training and data strategies on the three learning stages. We find several ways to better preserve the general preferring ability while training the customized RMs, especially general preference enrichment, and customized preference imitation learning. The DSP dataset and code are available at https://github.com/Linear95/DSP.
Giandomenico Cornacchia, Vito Walter Anelli, Fedelucio Narducci
et al.
Current AI regulations require discarding sensitive features (e.g., gender, race, religion) in the algorithm's decision-making process to prevent unfair outcomes. However, even without sensitive features in the training set, algorithms can persist in discrimination. Indeed, when sensitive features are omitted (fairness under unawareness), they could be inferred through non-linear relations with the so called proxy features. In this work, we propose a way to reveal the potential hidden bias of a machine learning model that can persist even when sensitive features are discarded. This study shows that it is possible to unveil whether the black-box predictor is still biased by exploiting counterfactual reasoning. In detail, when the predictor provides a negative classification outcome, our approach first builds counterfactual examples for a discriminated user category to obtain a positive outcome. Then, the same counterfactual samples feed an external classifier (that targets a sensitive feature) that reveals whether the modifications to the user characteristics needed for a positive outcome moved the individual to the non-discriminated group. When this occurs, it could be a warning sign for discriminatory behavior in the decision process. Furthermore, we leverage the deviation of counterfactuals from the original sample to determine which features are proxies of specific sensitive information. Our experiments show that, even if the model is trained without sensitive features, it often suffers discriminatory biases.
When the West African nation of Ghana attained its independence from colonial rule in 1957, its traditional culture was to be promoted in all sectors of public life. Similarly, what was construed as Ghanaian traditional religion was to be treated equally with Christianity and Islam. The ritual offering of libations to ancestral spirits and deities was considered the Ghanaian equivalent to Christian and Muslim prayers, and it has been performed side by side with them in all sorts of national events. Later on, the libation ritual became a symbol of both Ghana’s religious diversity and its national culture, transcending religious divisions. Many Christian groups, especially from the Pentecostal-charismatic movement, have refused to accept the public status of the libation ritual in view of its alleged immoral ‘pagan’ associations. When the pouring of libations was removed from the Independence Day ceremonies held at the state capital in 2011, the public debate soon turned to the relationship between the government and Pentecostal churches, and accusations of religious intolerance were levelled. This article discusses how the arguments about the status of the ritual boil down to differences in semiotic ideology and notions about proper agency – namely, how forms of agency pertaining to words, objects, persons and spiritual beings involved in the ritual are understood differently by the disputants.