Moral Responsibility and AI Extractivism: A STRUCTURAL ANALYSIS OF AI SUPPLY CHAINS

Author(s): Sharma, D. (2025)

Abstract:

AI ethics, despite its heterogeneous voices and perspectives, can fall short due to its principlistic 
orientation that only fixates on the technical aspects of the technology. This has resulted in shifts of 
theorising that not only acknowledge the technology’s sociotechnical embeddedness but also its structural 
aspects. With theorists hailing the structural turn in AI ethics, AI is being recognised as a global 
assemblage of natural resources, human labour and data from all around the world. However, the supply 
chains that funnel the resources for AI development result in a range of extractivist harms. And since 
these issues are beyond the boundaries of a single company entangled in a complex web made up of 
several supply chains, the problem of ‘many hands’ gets amplified on a global scale, yet has a 
disproportionate effect on the Majority World, with historical, complicating notions of moral 
accountability. Hence, the primary research question of this project is: how should one attribute moral 
responsibilities for the structural injustices involving extractivism across AI supply chains? To answer this 
question, this project borrows approaches from critical AI studies and philosophical literature on 
structural injustice, engaging in applied philosophy. After highlighting that the resources used in AI 
development are sourced through and uphold an extractive logic, I argue that AI ethics must acknowledge 
that the infrastructural harms, emerging from natural resource excavation, and exploitative harms, 
emerging from labour exploitation, are undivorced from the algorithmic harms as they are rooted in the 
same structure. I analyse these harms through the typologies of structural injustice outlined by Young 
(2011) and McKeown (2024b), to infer that natural resource extraction for AI development is a case of 
avoidable structural injustice, due to the foresseability of the harm of the mineral excavation and that 
labour exploitation for AI development is a case of deliberate structural injustice, as the structures that 
result in these harms are deliberately maintained. Furthermore, I argue that agents with power can be 
attributed moral responsibility for these structural injustices, as they can maintain, influence, and disrupt 
the structure that reproduces injustices. By tracing power through the supply chains and finding it 
concentrated at the downstream tier, I conclude that multinational corporations can be held morally 
responsible for the structural injustices involving extractivism across AI supply chains, due to the power 
they have in influencing and maintaining structures that result in injustice. Hence, positioning myself 
within the power-conscious structural turn in AI ethics, I argue for moral accountability by MNCs for the 
extractivist harms emerging from AI production and hint at the additional importance of the ordinary 
individual's political responsibility in bringing about structural change, also suggesting AI minimisation 
in terms of scale and necessity. 

Document(s):

Moral Responsibility and AI Extractivism.pdf