AI x Crisis: Tracing New Directions beyond Deployment and Use

Yang, Tianling et al. Book Part — Published Version AI x Crisis: Tracing New Directions beyond Deployment and Use Provided in Cooperation with: WZB Berlin Social Science Center Suggested Citation: Yang, Tianling et al. (2025) : AI x Crisis: Tracing New Directions beyond Deployment and Use, In: Association for Computing Machinery (ACM) (Ed.): Computing (X) Crisis : AR Adjunct '25: Adjunct Proceedings of the Sixth Decennial Aarhus Conference, ACM Digital Library, New York, NY, USA, pp. 1-4, https://doi.org/10.1145/3737609.3747096 , https://dl.acm.org/doi/proceedings/10.1145/3737609 This Version is available at: https://hdl.handle.net/10419/325702 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ AI x Crisis: Tracing New Directions beyond Deployment and Use Tianling Yang Technische Universität Berlin Berlin, Germany Weizenbaum Institute for the Networked Society Berlin, Germany [email protected] Srravya Chandhiramowuli Institute for Design Informatics University of Edinburgh Edinburgh, United Kingdom [email protected] Jana Heim WZB Berlin Social Science Center Berlin, Germany Weizenbaum Institute for the Networked Society Berlin, Germany [email protected] Camilla Salim Wagner Technische Universität Berlin Berlin, Germany Weizenbaum Institute for the Networked Society Berlin, Germany [email protected] Julian Posada Yale University New Haven, Connecticut, USA [email protected] Alex S Taylor Design Informatics University of Edinburgh Edinburgh , Scotland , United Kingdom [email protected] Rafael Grohmann University of Toronto Toronto, Ontario, Canada [email protected] Milagros Miceli Weizenbaum Institute for the Networked Society Technische Universität Berlin Berlin, Germany DAIR Institute Berlin, Germany [email protected] Abstract Faced with multiple, intersecting crises, numerous computing technologies have emerged and interacted with the crises. Amidst the growing prominence of AI, the discourses on AI-related harms predominantly focus on AI deployment and use, shifting attention away from their social and structural underpinnings. In response, this workshop seeks to reflect and map how AI intersects with the crises through framing the costs of AI. With costs of AI we refer to the human and natural toll of AI systems, such as labor exploitation, environmental degradation, and perpetuated social inequality, and emphasize the inherent and inevitable trade-offs in AI development and use. We invite contributions on various forms of AI-related costs, and critical engagement with methods to approach and address these costs. This workshop aims to (1) map the various costs of AI; (2) explore and reflect on concepts, frameworks, and methods to approach and engage with them; and (3) foster exchanges and collaborations in an interdisciplinary community. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs International 4.0 License. AAR Adjunct 2025, Aarhus N, Denmark ©2025 Copyright held by the owner/author(s). ACM ISBN 979-8-4007-1968-4/25/08 https://doi.org/10.1145/3737609.3747096 CCS Concepts •Human-centered computing → Human computer interaction (HCI);Collaborative and social computing. Keywords AI ethics, labor, environmental sustainability, AI harms, infrastructure ACM Reference Format: Tianling Yang, Srravya Chandhiramowuli, Jana Heim, Camilla Salim Wagner, Julian Posada, Alex S Taylor, Rafael Grohmann, and Milagros Miceli. 2025. AI x Crisis: Tracing New Directions beyond Deployment and Use. In Adjunct proceedings of the sixth decennial Aarhus conference: Computing X Crisis (AAR Adjunct 2025), August 18–22, 2025, Aarhus N, Denmark. ACM, New York, NY, USA, 4 pages. https://doi.org/10.1145/3737609.3747096 1 Introduction The past decade has borne witness to multiple, intersecting crises ranging from climate catastrophes to deepening socio-economic disparity and political polarization. Numerous computing technologies have been created within these crises, and some even in response to them. Nevertheless, many of them fuel the very crises they seek to alleviate, such as worsening disparities, aggravating climate impacts, and perpetuating extractive and colonial tendencies [ 6 ]. Further, the urgency and magnitude of these crises have been capitalized to justify the proliferation and adoption of scalable technological solutions, while undervaluing more locally and socially grounded alternatives. During this period of massive, uncritical AAR Adjunct 2025, August 18–22, 2025, Aarhus N, Denmark Yang, Chandhiramowuli, Heim, Wagner, Posada, Taylor, Grohmann and Miceli uptake of computing technologies, AI has been widely embraced and perhaps even celebrated as the poster child of technological innovation. Amidst its growing prominence, there is also increasing public concern over AI’s tendency to reproduce socio-historical biases [ 1 ] and deepen inequalities [ 8 ]. Much of this discourse centers harms arising from AI’s adoption in varied contexts and spurs redressal efforts focused on improving the design or performance of such systems. However, technical fixes and the underlying focus on model design are co-opted by narratives of inevitability, shifting attention away from the social and structural underpinnings of AI [3]. Reflecting on this decade marked by computing’s uncritical uptake and AI’s unchecked proliferation in the face of multiple, overlapping crises, we aim to engage with the varying, uneven impacts of AI by framing them as costs.Costs acknowledge the inherent and inevitable trade-offs in the development and use of AI systems, emphasizing the disproportionate burdens experienced in infrastructuring, improving, and maintaining AI and the need to account for and engage with various actors, especially those from the Majority World [ 22 ] that tend to be overlooked in WEIRD (i.e. Western, Educated, Industrialized, Rich, and Democratic) conceptions of AI ethics [ 21 ]. Costs thus offer a lens to foreground the multiple aspects of crises our worlds face and critically reflect and map how AI intersects with these crises — by addressing, reshaping, or deepening them. This workshop aims to recognize and trace the ways how AI intersects with crises and create space for collective and critical reflections. The workshop objectives include: (1) mapping the various costs of AI, (2) exploring and reflecting on concepts, frameworks, and methods to approach and address these costs, and (3) fostering exchange and collaboration in an interdisciplinary community that critically investigates the multi-faceted AI-related impacts. 2 Significance of the Workshop While predominant discourses center on harms in AI deployment and use, considering AI-related costs means exploring multiple ways and dimensions in which AI technologies are interwoven with crises. For instance, the development and use of AI technologies require large-scale human labor, such as in data generation [ 5 , 16 ], verification of automated outputs [ 26 ], and maintenance [ 7 ], leading to the extraction of data and labor globally [ 9 , 16 ]. Moreover, centering infrastructures of computing recognizes the materiality of computation and AI, including cables [ 24 ], data centers [ 11 ], and cloud infrastructures [ 18 ], and reveals the implications of tech giants’ dominance and control over critical AI infrastructure, such as economic centralization, smaller companies’ structural dependency on these infrastructures, and the potential of homogenization in knowledge production [ 4 , 15 ]. Additionally, despite being presented as a solution to combat climate changes, current AI systems are resource-intensive and have considerable environmental impacts regarding energy consumption [ 25 ], carbon emissions [ 19 ], water usage [ 14 ], lithium extraction and broader transformation of territory into resources and assets [ 12 ]. This raises an open question whether using AI technologies to combat climate catastrophe will create a larger environmental footprint overall [20]. Moreover, critical literature has pointed out that technical “fixes” to improve design or performance of AI systems can potentially shift attention away from the social and structural underpinnings of AI [ 17 ], whereas participatory methods may risk having shallow and even extractive community involvement, and masking hierarchical power structures within such settings [ 2 , 23 ]. Considering broader AI-related costs shows pathways to alternative methodologies, such as community-driven approaches [ 13 ], asset-based approaches [ 27 ] and action research [ 10 ]. These methodologies call for fostering changes that are grounded in social contexts and in networks of communities and stakeholders, such as building communities’ capacities, providing infrastructures and resources, and facilitating them to gain more agency. Aarhus 2025 conference centers on new perspectives and alternatives to discuss how computing has interacted with and transformed societies. We want to use this unique and timely opportunity to facilitate proactive discussions on different types of AI-related costs and re-imagine alternative ways to drive transformative change. 3 Topics of Interest In this workshop, we propose two guiding questions: What different types of costs of AI are there? How can research communities meaningfully engage with AI-related costs? We present four interconnected topics below and welcome further perspectives. • Human labor in AI production and use: What types of labor are integral to AI pipelines? What are their contexts, conditions, and characteristics? What types of costs arise from these labor practices? • Infrastructures of computing: What are the consequences of the increasing and structural dependence on AI-related infrastructure controlled by tech giants? • Environmental costs of AI: How can we measure, quantify, track, and visualize the environmental impacts the AI sector brings? How can we conceive of climate justice given the uneven distribution of benefits and environmental impacts in AI development and use? • Alternative methods to engage with AI-related costs: What are the theoretical frameworks, methodologies, and empirical cases to strengthen negotiation, resistance, and re-imagining of AI costs and futures? 4 Workshop Organization 4.1 Participant Recruitment We expect an audience size of around 15-20 participants (the organizers excluded), and encourage submissions from different professional backgrounds (e.g. academia, industry and NGOs) and various disciplines (e.g. critical computing, HCI, sociology, STS, infrastructure studies). We invite position papers (2-4 pages, excluding references) in ACM submission format. We plan to distribute the CfP on social media, specific mailing lists, and relevant Slack channels. 4.2 Workshop Activities and Tentative Schedule • 9:30 - 11:30: Welcome, Introduction and short paper presentations by participants. AI x Crisis: Tracing New Directions beyond Deployment and Use AAR Adjunct 2025, August 18–22, 2025, Aarhus N, Denmark •11:30 - 11:45: Coffee break • 11:45 - 12:30: Break-Out Group Discussion: Discussion in groups based on thematic focuses of position papers. This session will provide a space for the exchange of experiences, challenges, and constructive feedback. •12:30 - 13:45: Lunch • 13:45 - 15:00: Plenary Discussion: Mapping out various costs of AI, including relevant stakeholders, relationships, and contexts. •15:00 - 15:20: Coffee break • 15:20 - 16:50: Break-Out Group Discussion: Brainstorming and Method Reflection. Groups based on methods from participants’ position papers. Each group will choose a specific cost of AI, discuss how it can be addressed, and reflect on the strengths and limitations of particular methods and approaches. • 16:50 - 17:30: Wrap-Up: brief presentation of the discussions in the previous activity to the large group. •Optional: Group Dinner 5 Outcomes With the consent from the authors, we will publish a collection of submitted position papers as a workshop proceeding on ArXiv. This proceeding will also include a report summarizing key discussion points during the workshop. Moreover, we intend to facilitate networking and sustained collaboration among participants through a mailing list. Prior to the workshop, we will share a document with short bios of participants to get to know each other. Postconference, the mailing list will serve as a platform to discuss and share relevant events and publications. 6 Organizers Julian Posada is an Assistant Professor of American Studies at Yale University and a member of the Yale Law School’s Information Society Project and the Yale Institute for Foundations of Data Science. His research integrates theories and methods from information science, sociology, and human-computer interaction to examine how technology is developed and used within various historical, cultural, and social contexts. Camilla Salim Wagner is a research assistant at TU Berlin and the Weizenbaum Institute. As part of the Data Workers’ Inquiry team, she conducts community-based action-research to study the data work industry and support workers’ organization efforts. Tianling Yang is a doctoral researcher at TU Berlin and at the Weizenbaum Institute. Her research focuses on social and institutional contexts of the production of machine learning datasets, as well as actors and stakeholders in the global AI supply chains. Jana Heim is a doctoral researcher at the Weizenbaum Institute and WZB Berlin Social Science Center. Her research examines knowledge practices and digital technologies, particularly machine learning systems, and the social dynamics of their development and use. Srravya Chandhiramowuli examines the role of human values in data annotation and AI development. Her current research examines the work of data annotation for AI, to envision and inform just, equitable futures of AI. She is a PhD candidate in the Institute for Design Informatics at the University of Edinburgh. Alex Taylor has been contributing to Science & Technology Studies and Human-Computer Interaction (HCI) for over twenty years. His interests are in how digital technologies are co-constitutive of forms of knowing and doing, and, as a consequence, provide a basis for fundamental transformations in society. He is a Reader at Design Informatics, University of Edinburgh. Rafael Grohmann is an Assistant Professor of Media Studies at the University of Toronto Scarborough. He is the leader of DigiLabour initiative, the PI for Worker-Owned Intersectional Platforms (WOIP – an action research with delivery and tech workers in Brazil and Argentina), an editor for the journal Platforms and Society, and a member of Fairwork and Tierra Común. Milagros Miceli is head of the research group “Data, Algorithmic Systems and Ethics” at the Weizenbaum Institute and Research Lead at the Distributed AI Research Institute (DAIR). She leads the Data Workers’ Inquiry, a community-based research project and platform for data workers across the globe. References [1] Julia Angwin, Jeff Larson, Surya Mattu, Lauren Kirchner, and ProPublica. 2016. Machine Bias: There’s software used across the country to predict future criminals. And it’s biased against blacks. https://www.propublica.org/article/machinebias-risk-assessments-in-criminal-sentencing last accessed 2024/10/03. [2] Abeba Birhane, William Isaac, Vinodkumar Prabhakaran, Mark Diaz, Madeleine Clare Elish, Iason Gabriel, and Shakir Mohamed. 2022. Power to the People? Opportunities and Challenges for Participatory AI. In Proceedings of the 2nd ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization (Arlington, VA, USA) (EAAMO ’22). Association for Computing Machinery, New York, NY, USA, Article 6, 8 pages. https://doi.org/10.1145/ 3551624.3555290 [3] Abeba Birhane, Pratyusha Kalluri, Dallas Card, William Agnew, Ravit Dotan, and Michelle Bao. 2022. The Values Encoded in Machine Learning Research. In Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency (Seoul, Republic of Korea) (FAccT ’22). Association for Computing Machinery, New York, NY, USA, 173–184. https://doi.org/10.1145/3531146.3533083 [4] Sarah Burkhardt and Bernhard Rieder. 2024. Foundation models are platform models: Prompting and the political economy of AI. Big Data & Society 11, 2 (June 2024), 20539517241247839. https://doi.org/10.1177/20539517241247839 [5] Srravya Chandhiramowuli, Alex S. Taylor, Sara Heitlinger, and Ding Wang. 2024. Making Data Work Count. Proc. ACM Hum.-Comput. Interact. 8, CSCW1, Article 90 (apr 2024), 26 pages. https://doi.org/10.1145/3637367 [6] Nick Couldry and Ulises Ali Mejias. 2019. The costs of connection: how data is colonizing human life and appropriating it for capitalism. Stanford University Press, Stanford, [California]. [7] Sarah E. Fox, Samantha Shorey, Esther Y. Kang, Dominique Montiel Valle, and Estefania Rodriguez. 2023. Patchwork: The Hidden, Human Labor of AI Integration within Essential Work. Proc. ACM Hum.-Comput. Interact. 7, CSCW1, Article 81 (apr 2023), 20 pages. https://doi.org/10.1145/3579514 [8] Ana Cristina Bicharra Garcia, Marcio Gomes Pinto Garcia, and Roberto Rigobon. 2024. Algorithmic discrimination in the credit domain: what do we know about it? AI & SOCIETY 39, 4 (Aug. 2024), 2059–2098. https://doi.org/10.1007/s00146023-01676-3 [9] Mary L. Gray and Siddharth Suri. 2019. Ghost work: how to stop Silicon Valley from building a new global underclass. Houghton Mifflin Harcourt, Boston. [10] Gillian R. Hayes. 2011. The relationship of action research to human-computer interaction. ACM Trans. Comput.-Hum. Interact. 18, 3, Article 15 (Aug. 2011), 20 pages. https://doi.org/10.1145/1993060.1993065 [11] Mél Hogan. 2015. Data flows and water woes: The Utah Data Center. Big Data & Society 2, 2 (Dec. 2015), 205395171559242. https://doi.org/10.1177/ 2053951715592429 [12] Sebastián Lehuedé. 2022. Territories of data: ontological divergences in the growth of data infrastructure. Tapuya: Latin American Science, Technology and Society 5, 1 (Dec. 2022), 2035936. https://doi.org/10.1080/25729861.2022.2035936 [13] Chu Li, Katrina Oi Yau Ma, Michael Saugstad, Kie Fujii, Molly Delaney, Yochai Eisenberg, Delphine Labbé, Judy L Shanley, Devon Snyder, Florian P P Thomas, and Jon E. Froehlich. 2024. “I never realized sidewalks were a big deal”: A Case Study of a Community-Driven Sidewalk Accessibility Assessment using Project Sidewalk. In Proceedings of the 2024 CHI Conference on Human Factors in AAR Adjunct 2025, August 18–22, 2025, Aarhus N, Denmark Yang, Chandhiramowuli, Heim, Wagner, Posada, Taylor, Grohmann and Miceli Computing Systems (Honolulu, HI, USA) (CHI ’24). Association for Computing Machinery, New York, NY, USA, Article 969, 18 pages. https://doi.org/10.1145/ 3613904.3642003 [14] Pengfei Li, Jianyi Yang, Mohammad A. Islam, and Shaolei Ren. 2023. Making AI Less "Thirsty": Uncovering and Addressing the Secret Water Footprint of AI Models. http://arxiv.org/abs/2304.03271 arXiv:2304.03271 [cs]. [15] Dieuwertje Luitse. 2024. Platform power in AI: The evolution of cloud infrastructures in the political economy of artificial intelligence. Internet Policy Review 13, 2 (June 2024). https://doi.org/10.14763/2024.2.1768 [16] Milagros Miceli and Julian Posada. 2022. The Data-Production Dispositif. Proc. ACM Hum.-Comput. Interact. 6, CSCW2, Article 460 (Nov. 2022), 37 pages. https: //doi.org/10.1145/3555561 [17] Milagros Miceli, Julian Posada, and Tianling Yang. 2022. Studying Up Machine Learning Data: Why Talk About Bias When We Mean Power? Proc. ACM Hum.- Comput. Interact. 6, GROUP, Article 34 (Jan. 2022), 14 pages. https://doi.org/10. 1145/3492853 [18] Steven Gonzalez Monserrate. 2022. The Cloud Is Material: On the Environmental Impacts of Computation and Data Storage. MIT Case Studies in Social and Ethical Responsibilities of Computing Winter 2022 (jan 27 2022). https://mitserc.pubpub.org/pub/the-cloud-is-material. [19] David Patterson, Joseph Gonzalez, Quoc Le, Chen Liang, Lluis-Miquel Munguia, Daniel Rothchild, David So, Maud Texier, and Jeff Dean. 2021. Carbon Emissions and Large Neural Network Training. arXiv:2104.10350 [cs.LG] https://arxiv.org/ abs/2104.10350 [20] Rainer Rehak. 2023. Artificial Intelligence for Real Sustainability? In Shaping digital transformation for a sustainable society: contributions from Bits & Bäume. Technische Universität Berlin, 26–31. https://depositonce.tu-berlin.de/handle/ 11303/18717 Publisher: Technische Universität Berlin. [21] Ali Akbar Septiandri, Marios Constantinides, Mohammad Tahaei, and Daniele Quercia. 2023. WEIRD FAccTs: How Western, Educated, Industrialized, Rich, and Democratic is FAccT?. In Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency (Chicago, IL, USA) (FAccT ’23). Association for Computing Machinery, New York, NY, USA, 160–171. https://doi.org/10.1145/ 3593013.3593985 [22] Ranjit Singh. 2024. Ordinary Ethics of Governing AI: Artificial intelligence has real impacts on the everyday lives of people all around the world. These stories invite a broader conversation on research and policy about AI in the global south. https://carnegieendowment.org/research/2024/04/ordinary-ethics-ofgoverning-ai?lang=en last accessed 2024/10/03. [23] Mona Sloane, Emanuel Moss, Olaitan Awomolo, and Laura Forlano. 2022. Participation Is not a Design Fix for Machine Learning. In Equity and Access in Algorithms, Mechanisms, and Optimization. ACM, Arlington VA USA, 1–6. https://doi.org/10.1145/3551624.3555285 [24] Nicole Starosielski. 2015. The undersea network. Duke University Press, Durham. [25] Emma Strubell, Ananya Ganesh, and Andrew McCallum. 2019. Energy and Policy Considerations for Deep Learning in NLP. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Anna Korhonen, David Traum, and Lluís Màrquez (Eds.). Association for Computational Linguistics, Florence, Italy, 3645–3650. https://doi.org/10.18653/v1/P19-1355 [26] Paola Tubaro, Antonio A Casilli, and Marion Coville. 2020. The trainer, the verifier, the imitator: Three ways in which human platform workers support artificial intelligence. Big Data & Society 7, 1 (Jan. 2020), 205395172091977. https://doi.org/10.1177/2053951720919776 [27] Marisol Wong-Villacres, Aakash Gautam, Deborah Tatar, and Betsy DiSalvo. 2021. Reflections on Assets-Based Design: A Journey Towards A Collective of AssetsBased Thinkers. Proc. ACM Hum.-Comput. Interact. 5, CSCW2, Article 401 (Oct. 2021), 32 pages. https://doi.org/10.1145/3479545