AI & Society最新文献

{"title":"Capability determinism, energy and AI labour substitution","authors":"Will Mbioh","doi":"10.1007/s00146-026-03251-y","DOIUrl":"10.1007/s00146-026-03251-y","url":null,"abstract":"<div><p>Contemporary policy and media discourse on AI and labour follows a recurring pattern. A demonstration of model capability runs through an inference about workplace substitution to a distributional politics, with the variables that would decide the outcome treated as background. This paper names that pattern capability determinism situates it within Wyatt’s account of soft technological determinism, and reframes the substitution question as a unit-cost economics test. On the AI side, cost per task decomposes into a variable inference cost dominated by electricity and the supply chain that produces compute, an amortised integration cost recoverable across volume, and a residual supervision cost that regulatory and professional regimes will not let fall to zero. Once these three lines are loaded, the substitution discourse turns out to operate as a feedback structure: capability gain stresses the supporting systems faster than it improves the calculus on which substitution depends. The variables outside the AI debate, energy markets, integration economics, professional regulation, capital-market patience, and the geopolitics of compute supply chains decide whether substitution clears in any given place at any given time. Substitution arrives selectively, configurationally, and at a more modest scale than the dominant discourse implies.</p></div>","PeriodicalId":47165,"journal":{"name":"AI & Society","volume":"41 10","pages":"9297 - 9308"},"PeriodicalIF":6.1,"publicationDate":"2026-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s00146-026-03251-y.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230799","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Epistemic decolonisation and the integration of African traditional medicine: toward an epistemic redress model for inclusive AI in mental healthcare","authors":"Linda Maqutu","doi":"10.1007/s00146-026-03182-8","DOIUrl":"10.1007/s00146-026-03182-8","url":null,"abstract":"<div><p>Artificial intelligence (AI), particularly machine learning (ML) systems, is increasingly used in mental healthcare to support diagnosis and treatment. These systems analyse large datasets—including speech patterns, social media activity, and biometric indicators—to detect conditions such as depression and anxiety. Although often framed as objective and clinically neutral, these technologies embed epistemic and cultural assumptions derived largely from Global North contexts. This paper advances a decolonial critique of AI in mental healthcare by arguing that ML systems reproduce colonial epistemologies through their underlying data practices, categories, and validation structures. Drawing on Fricker’s (2007) account of epistemic injustice and Dotson’s (2012) theory of epistemic oppression, I argue that the exclusion of Indigenous knowledge systems (IKS)—particularly African traditional medicine (ATM)—from AI design and implementation constitutes both testimonial and hermeneutical injustice. Building on this critique, the paper develops a preliminary Epistemic Redress Model for Mental Health AI (Sect. 5), which outlines the epistemic conditions required for more inclusive and context-sensitive machine learning systems. Using South Africa as a case study, I demonstrate how ATM offers alternative conceptual and therapeutic resources for understanding mental illness that are systematically marginalised within dominant biomedical and data-driven paradigms. The paper concludes that without structured epistemic redress, AI systems in mental healthcare risk reinforcing colonial hierarchies of knowledge and producing ethically and clinically inadequate models of care.</p></div>","PeriodicalId":47165,"journal":{"name":"AI & Society","volume":"41 10","pages":"9121 - 9137"},"PeriodicalIF":6.1,"publicationDate":"2026-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s00146-026-03182-8.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230651","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Professional ethics and practical wisdom in AI-mediated professional environments","authors":"Henrique Cardoso","doi":"10.1007/s00146-026-03204-5","DOIUrl":"10.1007/s00146-026-03204-5","url":null,"abstract":"<div><p>The growing incorporation of artificial intelligence into professional environments has intensified debates about how ethical judgement is formed and sustained under technological mediation. This article examines these issues through the case of VICTOR, a machine learning system developed by Brazil’s Supreme Federal Tribunal to classify extraordinary appeals according to constitutional themes of general repercussion, and through the later expansion of tools such as VitórIA. The analysis distinguishes between digital procedural infrastructures and machine learning systems to show that the ethical significance of AI does not depend only on whether such systems make final decisions. Even when judicial authority remains formally human, algorithmic classification can reshape the informational conditions under which professional judgement is exercised. Drawing on moral development theory, virtue ethics, moral identity research and debates on moral deskilling and interpretive uncertainty, the article argues that AI-mediated professional environments create a specific obligation of critical oversight. It concludes that the ethical challenge of AI lies not simply in automation, but in the transformation of the conditions under which practical wisdom, professional authority and ethical autonomy are sustained.</p></div>","PeriodicalId":47165,"journal":{"name":"AI & Society","volume":"41 10","pages":"9157 - 9168"},"PeriodicalIF":6.1,"publicationDate":"2026-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230797","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Revising research practices for singing data collection","authors":"Kelsey Cotton,&nbsp;André Holzapfel,&nbsp;Katja de Vries,&nbsp;Karl Berglund,&nbsp;Kıvanç Tatar","doi":"10.1007/s00146-026-03205-4","DOIUrl":"10.1007/s00146-026-03205-4","url":null,"abstract":"<div><p>As AI voice synthesis enables increasingly sophisticated vocal deepfakes and non-consensual voice cloning, the governance, licensing and access of singing datasets has become an urgent concern for data-contributors, who face significant harms from downstream and non-consensual usage of their singing data. Singing datasets are foundational to the development of high fidelity voice AI synthesis, yet current data collection practices pose challenges: data-contributors have an event-centric contribution to datasets which constrains the scope of the power they have in influencing dataset licensing and access decisions; and they face greater potential harms from non-consensual downstream use of their data. This is in contrast to data-collectors, whose involvement with the dataset directly concerns licensing and access decisions, and who do not risk the same harms as data-contributors. Current singing data collection practices suggest an imbalance in the power-to-interest stakes held by data-contributors relative to data-collectors. To investigate the power-to-interest differences between data-contributors and data-collectors, we apply the Ethically Aligned Stakeholder Elicitation (EASE) framework to three singing datasets, mapping the power-to-interest stakes of data-contributors and considering these stakes against their exposure to potential harms from downstream data usage. Our analysis consistently demonstrates that data-contributors occupy lower-power positions compared with data-collectors; they hold minimal decision-making authority over how their data may be accessed; and they face greater risk of harm from non-consensual downstream use of their data. A persistent temporal symmetry emerges: while data-contributors’ involvement in datasets is event-centric, their vulnerability to harm is temporally unbounded, extending beyond the moment of contribution as new AI capabilities develop. Drawing from a cross-section of data ethics, decolonial data perspectives and artificial intelligence legislation, we propose revised research guidelines which reposition data-contributors more centrally in dataset governance and licensing decisions. We contribute with three consideration factors to better consider data-contributors’ interests: stakeholder authority in dataset licensing, use and access decisions; the temporal scope of stakeholder involvement with the dataset; and vulnerability to harm from downstream data usage. Our findings highlight an urgent need for researchers to consider what licensing-related steps could be taken to protect the personality or identity rights of human participants within an AI climate in which singing data is increasingly treated as a freely available commodity for downstream use.</p></div>","PeriodicalId":47165,"journal":{"name":"AI & Society","volume":"41 10","pages":"9617 - 9644"},"PeriodicalIF":6.1,"publicationDate":"2026-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s00146-026-03205-4.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230648","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"AI and actor-specific decisions","authors":"Teppo Felin,&nbsp;Mari Sako,&nbsp;Jessica Hullman","doi":"10.1007/s00146-026-03202-7","DOIUrl":"10.1007/s00146-026-03202-7","url":null,"abstract":"<div><p>Artificial intelligence (AI) is increasingly seen as potentially replacing humans in decision-making and problem-solving across many domains. AI is effective for many well-specified decisions. But we argue that AI cannot deal with what we call “actor-specificity.” Actor-specific decisions and problems are (a) forward-looking, (b) individual and idiosyncratic, (c) reasoning-intensive, and (d) experimental—requiring intervention in the world to facilitate “counter-to-data” reasoning. These four criteria, captured by the “FIRE” acronym, function as exclusion criteria: they identify when decisions should not be delegated to AI. These criteria are unified by an overarching concept we call “actor-specificity”: the recognition that such decisions cannot be separated from the particular agent—their beliefs, goals, and theories. The “actor” in actor-specificity refers to the focal decision maker, highlighting the need for a <i>first-person point of view in decision-making</i>—an approach that cannot be modeled from the third-person, population-level perspective that is the basis of AI. As such, the FIRE criteria offer an epistemic and normative basis for determining which decisions should remain with human actors, distinct from the current and future technical capabilities of AI.</p></div>","PeriodicalId":47165,"journal":{"name":"AI & Society","volume":"41 10","pages":"9597 - 9615"},"PeriodicalIF":6.1,"publicationDate":"2026-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s00146-026-03202-7.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230673","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Toward an integrated multilevel framework for AI moral literacy: bridging individual and organizational ethical capacities","authors":"Jerome Cleofas","doi":"10.1007/s00146-026-03099-2","DOIUrl":"10.1007/s00146-026-03099-2","url":null,"abstract":"<div><p>As artificial intelligence (AI) systems increasingly shape lives across various sectors of society, a heightened moral imperative has emerged: individuals and organizations must be equipped with ethical capacities to responsibly navigate an AI-driven future. This paper fowards an <i>integrative multilevel framework of AI moral literacy</i>, which transcends cognitive–affective abilities and technological skills and emphasizes relational, context-sensitive ethical capacity. Drawing from frameworks on individual moral literacy and organizational moral learning, this discussion paper integrates these perspectives to conceptualize AI moral literacy as a recursive, co-constitutive dynamic between people and institutions. Through this multilevel integration, the emerging dialectical framework attends to both personal moral agency and the structural conditions that enable or constrain it. Elaboration of intersections between dimensions of individual and organzational moral literacy in the context of AI are also presented.</p></div>","PeriodicalId":47165,"journal":{"name":"AI & Society","volume":"41 on","pages":"7509 - 7521"},"PeriodicalIF":6.1,"publicationDate":"2026-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148807550","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ArtIA: imagining a new digital commons for creative AI","authors":"Maurice Jones,&nbsp;Fenwick McKelvey,&nbsp;Bart Simon,&nbsp;Marek Blottière","doi":"10.1007/s00146-026-03192-6","DOIUrl":"10.1007/s00146-026-03192-6","url":null,"abstract":"<div><p>While American screenwriters were resisting AI capture, a vibrant community of digital creatives and arts organizations across Québec and Canada formed the alliance ArtIA. ArtIA advances commons-based alternatives to cultural AI, in response to the observation that while AI is rapidly transforming creative tools and workflows, the cultural sector lacks the critical resources, reference frameworks, support systems, and training needed to navigate this shift. The AI commons is conceptually contested, as corporate and governmental discourses increasingly frame AI as a “common good,” while probabilistic AI systems, common training datasets, and debates over AI as public infrastructure generate a muddled appropriation of the commons. ArtIA advances commoning as a situated practice grounding visions of the future in cultural needs, sectoral challenges, and traditions of commoning in Québec and Canada. This paper traces the 18-month stakeholder consultation organized by ArtIA, which enrolled 204 participants and collectively aimed to imagine and develop concrete interventions toward an AI commons situated at the intersection of Canadian AI governance as industrial policy and the localized practices of digital creation. Findings grounded in situational analysis and action research show the enrollment of four interrelated actor groups—cultural organizations, digital creatives, commons networks, and researchers. These actors formed new alliances and produced three concrete imaginaries for a creative AI commons: (1) Artisanal AI, (2) Common Compute, and (3) Critical AI Literacies. In closing, we argue that technologies developed and governed at the right scale—accessible, situated, and collectively stewarded—can operate as material levers for social solidarity, collective agency and shared responsibility that ideological politics and industrial policies cannot.</p></div>","PeriodicalId":47165,"journal":{"name":"AI & Society","volume":"41 10","pages":"9505 - 9520"},"PeriodicalIF":6.1,"publicationDate":"2026-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230767","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Co-designing AI systems with value-sensitive citizen science","authors":"Sachit Mahajan,&nbsp;Dirk Helbing","doi":"10.1007/s00146-026-03174-8","DOIUrl":"10.1007/s00146-026-03174-8","url":null,"abstract":"<div><p>As AI systems increasingly influence everyday life, integrating diverse community values is both ethically essential and practically urgent. This paper presents Value-Sensitive Citizen Science (VSCS)—a systematic framework that combines Value-Sensitive Design (VSD) with citizen science to support meaningful public participation in AI development. VSCS addresses gaps in the existing approaches by integrating culturally grounded methods and cognitive scaffolding through the Participatory Value-Cognition Taxonomy (PVCT). Community members engage as co-researchers through iterative cycles guided by extended scenario reasoning (What-if, If-then, Then-what, What-now), translating local values into actionable technical requirements. The framework also embeds governance mechanisms to ensure adaptability, accountability, and ongoing oversight throughout the AI lifecycle. By bridging participatory design with algorithmic accountability, VSCS challenges monocultural and top–down approaches to AI. We discuss practical implications, including power asymmetries, scalability, and epistemic justice, and propose strategies for policymakers and practitioners seeking to advance inclusive, value-driven AI design across diverse sociotechnical contexts.</p></div>","PeriodicalId":47165,"journal":{"name":"AI & Society","volume":"41 on","pages":"7267 - 7282"},"PeriodicalIF":6.1,"publicationDate":"2026-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s00146-026-03174-8.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148807446","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The replacement-augmentation paradox: techno-perceptual dynamics of AI adoption in medical imaging and the future of work","authors":"Syed Mohammad Waheed","doi":"10.1007/s00146-026-03145-z","DOIUrl":"10.1007/s00146-026-03145-z","url":null,"abstract":"<div><p>The implementation of artificial intelligence (AI) in radiology has garnered significant attention across both research and practice over the last decade. However, opinions on whether AI will augment or replace radiologists in their work have so far been divided, and the debate around this continues to linger. This article conceptually structures and terms this phenomenon as the “replacement-augmentation paradox” of AI in radiology—the persistent coexistence of replacement fears and augmentation perceptions within the same professional community. Rather than treating these positions as mutually exclusive or empirically resolvable through performance metrics alone, this article conceptualises the replacement-augmentation paradox as a synthesising framework that structures and interprets divergent, evidence-informed narratives of AI integration in radiology. Drawing predominantly on sociotechnical and perception-based research, this article juxtaposes the two coexistent narratives and synthesises collective observations to shed light on relatively less emphasised aspects shaping replacement-augmentation debates. The findings obtained here explicate the implementation of AI tools (such as deep learning) for medical imaging and early diagnosis in radiology through a proposed “techno-perceptual” framework where the augmentation and replacement of the existing work practices are both plausible outcomes (O), subject to preconditions. Alleviation of fears and determination of the resultant narrative is thus linked with the provision of three jointly enabling conditions: training (T), evidence (E), and decisional control (C), the presence or absence of which are deemed to be directly consequential in terms of yielding perceptions of augmentation or replacement, respectively. The importance and interplay of these factors are delineated here through the novel Boolean equation O = T ∧ E ∧ C, a corresponding logical table, a process flow diagram, and a radiologist typology matrix depicting four professional archetypes.</p></div>","PeriodicalId":47165,"journal":{"name":"AI & Society","volume":"41 9","pages":"8481 - 8493"},"PeriodicalIF":6.1,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149098258","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Imaging urban doppelgängers: art and the digital twin in South Korea","authors":"Melanie Wilmink","doi":"10.1007/s00146-026-03191-7","DOIUrl":"10.1007/s00146-026-03191-7","url":null,"abstract":"<div><p>This article examines how South Korean artists engage the concept and material infrastructures of the digital twin. While often dismissed as decorative or superfluous, artistic images of the city are an essential part of place-making, as a counter-point to the usually prioritized imagination of municipal bureaucrats, private developers, architects, and urban planners. Artworks structure a hybrid phenomenology that positions the spectator as simultaneously “here” in the physical and “there” in the virtual or conceptual; this doubled spectatorship reveals the constructed frames of experience, thereby creating an uncanny form of attention that implicates the spectator in a relationship with the image. By visualizing the material world and its digital double side by side, Hoonida Kim draws attention to new forms of human perception engendered by digital twin tools. Turning from the hardware to data curation, members of Red Onion collective (Kim et al. 2020) highlight the flawed or narrow applications of urban “big data” through the City of Seoul’s official Open Data portal. Finally, the artist-activist group Listen to the City mirror the data-collection and analytic methodologies of the smart city through an artistic, feminist, and community-driven lens in order to resist gentrification. With the normalization of digital twins and similar “smart” technologies in urban environments, it is critical to bring the invisible components of urban infrastructure to the foreground in order to examine their effects on the material, conceptual, and societal spaces of the city. By leveraging artistic imagination and image-production, these artists expand on the existing technical and social discourse of the digital twin and the smart city.</p></div>","PeriodicalId":47165,"journal":{"name":"AI & Society","volume":"41 10","pages":"9545 - 9555"},"PeriodicalIF":6.1,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230647","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}