From Technological Neutrality to Value Embedding: Power Structures and Discursive Contestation in Artificial Intelligence Standard Setting

Abstract

The assumption of technological neutrality has long underpinned technical standard setting, suggesting that standards merely reflect objective, value-free technical specifications. This assumption is increasingly untenable in artificial intelligence (AI) governance, where standards function as de facto regulatory instruments embedding specific values, interests, and worldviews. This study challenges the technological neutrality thesis by examining power structures and discursive contestation in AI standard setting processes. Through four comparative case studies analyzing the European Union AI Act standardization request, the IEEE 7000 value embedding framework, NIST's zero drafts pilot project, and international standard negotiations within ISO/IEC JTC 1/SC 42, this research investigates how different actors exercise power to shape AI standards. Drawing on publicly available policy documents, technical reports, and standardization records, the study employs qualitative document analysis to map power asymmetries and discursive strategies. Findings reveal that AI standard setting constitutes a contested political arena where value embedding occurs through systematic power asymmetries: technical expertise serves as a barrier to participation, private actors exercise disproportionate influence, and competing value frameworks reflect geopolitical tensions. Ultimately, the study contributes to critical AI governance scholarship by exposing the inherently political nature of technical standardization. It concludes by proposing actionable pathways for more inclusive, democratically accountable standard setting processes that can better accommodate diverse societal needs and ethical considerations.