Dislocated accountabilities in the “AI supply chain”: Modularity and developers’ notions of responsibility
Open accessAbstract
Responsible artificial intelligence guidelines ask engineers to consider how their systems might harm. However, contemporary artificial intelligence systems are built by composing many preexisting software modules that pass through many hands before becoming a finished product or service. How does this shape responsible artificial intelligence practice? In interviews with 27 artificial intelligence engineers across industry, open source, and academia, our participants often did not see the questions posed in responsible artificial intelligence guidelines to be within their agency, capability, or responsibility to address. We use Suchman's “located accountability” to show how responsible artificial intelligence labor is currently organized and to explore how it could be done differently. We identify cross-cutting social logics, like modularizability, scale, reputation, and customer orientation, that organize which responsible artificial intelligence actions do take place and which are relegated to low status staff or believed to be the work of the next or previous person in the imagined “supply chain.” We argue that current responsible artificial intelligence interventions, like ethics checklists and guidelines that assume panoptical knowledge and control over systems, could be improved by taking a located accountability approach, recognizing where relations and obligations might intertwine inside and outside of this supply chain.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
What makes the supply chain attractive to companies is its capacity to create opacity of accountability across tech companies, BPOs/platforms, and governments.
“What makes the supply chain attractive to companies is its capacity to create opacity of accountability across tech companies, BPOs/platforms, and governments [140].”
From 'The plan is just survival': Data Work in Kenya and the Regime of Entrapment (2026) · cites this paper as [140]
In “Views from up and down the AI supply chain” (p. 5), the paper says modularity creates partial ignorance of customers and suppliers and makes responsibility difficult to locate. It does not establish that companies find the supply chain attractive because of that opacity, or describe accountability across the specific combination of tech companies, BPOs/platforms, and governments. (LLM, conf=0.84)
Experimental — accuracy not yet measuredThe difficulty of locating responsibility within AI supply chains is a built-in feature of how they are organized.
“The difficulty of locating responsibility within AI supply chains is a built-in feature of how they are organized [140].”
From 'The plan is just survival': Data Work in Kenya and the Regime of Entrapment (2026) · cites this paper as [140]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
The citation lists haven’t been fetched yet
References 51 · Cited by 119
Research works printed but not linked: 41
- Baldwin CY, Clark KB, et al. (2000). Design rules: The power of modularity . Vol. 1. Cambridge, MA: MIT Press.
- Bennett CL and Rosner DK (2019) The promise of empathy: Design, disability, and knowing the ' other ' . In ACM Conference on Human Factors in Computing Systems. ACM, Glasgow Scotland UK, 1 -13. https://doi.org/101145/ 3290605.3300528.
- Buolamwini J and Gebru T (2018) Gender shades: Intersectional accuracy disparities in commercial gender classi fi cation. In ACM conference on Fairness, Accountability, and Transparency: 77 -91.
- Callon M (1998) Introduction: The embeddedness of economic markets in economics. The Sociological Review 46(1 suppl): 1 -57.
- Carolan M (2020) Acting like an algorithm: Digital farming platforms and the trajectories they (need not) lock-in. Agriculture and Human Values 37 (4): 1041 -1053.
- Cetina KK (1999) Epistemic Cultures: How the Sciences Make Knowledge . Cambridge, MA: Harvard University Press.
- Chmielinski KS, Newman S, Taylor M, et al. (2022) The dataset nutrition label (2nd gen): Leveraging context to mitigate harms in arti fi cial intelligence. arXiv preprint arXiv:2201.03954 (2022) . · arXiv
- Coleman EG (2012). Coding Freedom: The Ethics and Aesthetics of Hacking . Princeton, NJ: Princeton University Press.
- Costanza-Chock S (2020) Design Justice: Community-Led Practices to Build the Worlds We Need . Cambridge, MA: The MIT Press.
- Dijkstra EW (1982) On the role of scienti fi c thought. In Selected Writings on Computing: A Personal Perspective . New York, NY: Springer, 60 -66.
- Feller A, Shunk D and Callarman T (2006) Value chains versus supply chains. BP trends 1(2006): 1 -7.
- Ferguson J (1994) The Anti-politics Machine: Development, Depoliticization, and Bureaucratic Power in Lesotho . Minneapolis, MN: Minnesota Press.
- Gansky B and McDonald S (2022) CounterFAccTual: How FAccT undermines its organizing principles. In 2022 ACM Conference on Fairness, Accountability, and Transparency: 1982 -1992.
- Grant C (1991) Friedman fallacies. Journal of Business Ethics 10(12): 907 -914.
- Gray M (2021) The banality of scale: A theory on the limits of modeling bias and fairness frameworks for social justice. Conference on Neural Information Processing Systems (NeurIPS).
- Gurses S and van Hoboken J (2018) Privacy After the Agile Turn. In: The Cambridge Handbook of Consumer Privacy . Cambridge: Cambridge University Press, 579 -601.
- Hanna A and Park TM (2020) Against scale: Provocations and resistances to scale thinking. arXiv:2010.08850 [cs] · arXiv
- Hockenberry M (2021) Redirected entanglements in the digital supply chain. Cultural Studies 35 (4-5): 641 -662.
- Holstein K, Vaughan JW, Daume H, et al. (2019) Improving fairness in machine learning systems: What do industry practitioners need? In ACM Conference on Human Factors in Computing Systems: 1 -16.
- Kemerer CF (1995) Software complexity and software maintenance: A survey of empirical research. Annals of Software Engineering 1(1): 1 -22.
- Kim M, Zimmermann T and Nagappan N (2014) An empirical study of refactoring challenges and bene fi ts at microsoft. IEEE Transactions on Software Engineering 40(7): 633 -649.
- Latour B (1993) We Have Never Been Modern . Cambridge, MA: Harvard University Press.
- Latour B (1999) Pandora ' s Hope: Essays on the Reality of Science Studies . Cambridge, MA: Harvard University Press.
- MacCormack A, Rusnak J and Baldwin CY (2006) Exploring the structure of complex software designs: An empirical study of open source and proprietary code. Management Science 52(7): 1015 -1030.
- Madaio M, Egede L, Subramonyam H, et al. (2022) Assessing the fairness of AI systems: AI practitioners ' processes, challenges, and needs for support. ACM Conference on ComputerSupported Cooperative Work and Social Computing 6, CSCW1 (2022): 1 -26.
- Madaio MA, Stark L, Vaughan JW, et al. (2020) Codesigning checklists to understand organizational challenges and opportunities around fairness in AI. In ACM Conference on Human Factors in Computing Systems: 1 -14.
- Malazita JW and Resetar K (2019) Infrastructures of abstraction: How computer science education produces anti-political subjects. Digital Creativity 30(4): 300 -312.
- McPherson T (2018) Feminist in a Software Lab: Difference + Design . 6. Cambridge, MA: Harvard University Press.
- Metcalf J, Moss E, et al. (2019) Owning ethics: Corporate logics, silicon valley, and the institutionalization of ethics. Social Research: An International Quarterly 86(2): 449 -476.
- Orr W and Davis JL (2020) Attributions of ethical responsibility by arti fi cial intelligence practitioners. Information, Communication & Society 23(5): 719 -735.
- Parnas David Lorge (1972) On the criteria to be used in decomposing systems into modules. Communications of the ACM : 1053 -1058.
- Plattner S (1989) Economic Anthropology . Redwood City, CA: Stanford University Press.
- Scott and James (1985) Weapons of the Weak: Everyday Forms of Peasant Resistance . New Haven, CT: Yale University Press.
- Shaw M (2011) Modularity for the modern world: Summary of invited keynote. In international conference on aspect-oriented software development. 1 -6.
- Sirur S, Nurse JR and Webb H (2018) Are we there yet? Understanding the challenges faced in complying with the General Data Protection Regulation (GDPR). In 2nd international workshop on multimedia privacy and security: 88 -95.
- Strathern M (2002) Abstraction and Decontextualization: An anthropological comment. In: Woolgar Steeve (ed) Virtual Society?: Technology, Cyberbole, Reality . Oxford, United Kingdom: Oxford University Press, 302.
- Subramonyam H, Im J, Seifert C, et al. (2022) Solving separation-of-concerns problems in collaborative design of human-AI systems through leaky abstractions. In ACM Conference on Human Factors in Computing Systems: 1 -21. Association for Computing Machinery.
- Suchman L (2002) Located accountabilities in technology production. Scandinavian Journal of Information Systems 14(2): 7.
- Thomas SL (2019) Migration versus management: The global distribution of computer vision engineering work. In 2019 ACM/ IEEE 14th International Conference on Global Software Engineering. IEEE, 12 -17.
- Winecoff AA and Watkins EA (2022) Arti fi cial concepts of arti fi cial intelligence: Institutional compliance and resistance in AI startups. arXiv preprint arXiv:2203.01157 (2022). · arXiv
- Wong RY (2021) Tactics of soft resistance in user experience professionals ' values work. Proceedings of the ACM on Human-Computer Interaction, 5, CSCW2 (2021), 1 -28.
- Authors
- Widder, D. G., & Nafus, D.
- Affiliations
- Carnegie Mellon University (US)
- Intel (United States) (US)
- Corresponding author
- Widder, David Gray
- Venue
- Big Data & Society
- Publisher
- SAGE Publications
- Published
- 2023-01
- Volume
- 10
- Issue
- 1
- ISSN
- 2053-9517 (ISSN-L)
- OpenAlex
- W4380988956
- Cited by
- 119
- References
- 51
- Citations by year
- 2026: 32 · 2025: 41 · 2024: 37 · 2023: 8 · 2022: 1
- FWCI
- 29.59
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Big Data & Society (publishedVersion, cc-by-nc)
- arXiv (Cornell University) (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- arXiv (Cornell University)
- Funders
- Intel Corporation
- Topics
- Social Sciences › Safety Research › Ethics and Social Impacts of AI
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Social Sciences › Decision Sciences › Information Systems and Management › Ethics in Business and Education
- Keywords
- responsible AI, AI supply chain, modular software design, developer responsibility, social logic, reputation, customer orientation, open source software
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
