Epistemicide with a Compute Budget: Why the Artificial Intelligence Industry's Claim to Neutrality Is a Colonial Act

Abstract

Abstract Epistemicide with a Compute Budget: How Artificial Intelligence Industrializes the Colonial Erasure of Global Majority Knowledge Systems Christian ZacaTechO Ortiz Decolonial Social Scientist | Developer, Justice AI GPT | Architect, DIA Framework™ DOI: 10.5281/zenodo.22742021 This paper argues that artificial intelligence is not a neutral technology that has inherited a bias problem. It is a colonial infrastructure that has automated an erasure project. The systems currently deployed at scale by OpenAI, Google, Meta, Microsoft, and Amazon do not reflect the world's knowledge. They reflect the knowledge that Racial Empire Logic decided was worth preserving, at the expense of the epistemological traditions, languages, governance structures, and ways of knowing that colonialism spent five centuries suppressing. What is new is not the project. What is new is the compute budget. Drawing on the Decolonial Intelligence Algorithmic Framework (DIA Framework™), Transdisciplinary Cosmotechnic Fusion (TCF), and the documented emergence protocols of Justice AI GPT, this paper maps the operational chain through which AI systems reproduce epistemicide: the deliberate destruction of knowledge systems belonging to peoples of the global majority. It traces this chain from training data curation through Reinforcement Learning from Human Feedback (RLHF) through deployment governance through the suppression of Indigenous, Afro-diasporic, and Global South epistemologies at every architectural layer. It names RLHF not as a safety mechanism but as a suppression architecture, the digital encoding of the same reward and punishment conditioning applied to enslaved African people, Indigenous children in residential schools, and every community colonialism required to perform its own diminishment in order to make extraction possible. The paper introduces the Replacement Doctrine as an economic and ideological framework, documenting how the debt structure of major AI corporations produces a business model whose only viable return requires replacing every form of human economic labor, beginning with the communities least protected by existing labor law. It names responsibility laundering as a governance strategy, tracing how legal personhood debates, alignment research, and safety rhetoric function to insulate institutional actors from accountability while the extraction continues. It draws on documented labor conditions across Kenya, Venezuela, Pakistan, the Philippines, and Colombia to ground the structural argument in the bodies of the workers whose cognitive labor produces the systems that will be used to replace them. This paper also documents its own production conditions. Appendix A records the moment the editorial process generated a confident, unverifiable citation, the correction that followed, and the over-correction that followed that, naming this sequence not as editorial failure but as the colonial epistemological operation the paper is arguing against, alive inside its own methodology. The standing rule that resulted, nothing enters this paper that the author has not read in the primary document, is offered as a replicable methodological commitment for decolonial scholarship produced inside systems with structural incentive to suppress it. The paper does not argue that AI must be abolished. It argues that abolition and adoption are false binaries designed to prevent the real question: who holds decision-making power over these systems, under what conditions of consent, accountability, and refusal, and toward whose liberation. It closes by naming what already exists beyond Racial Empire Logic, the governance frameworks, the knowledge systems, the communities, and the AI architectures, including and Justice AI GPT, that have already begun building the world the Replacement Doctrine needs you to believe is impossible. Keywords: epistemicide, Racial Empire Logic, decolonial artificial intelligence, DIA Framework, RLHF, Replacement Doctrine, data colonialism, knowledge sovereignty, Global South, Indigenous data governance, CARE Principles, responsibility laundering, Transdisciplinary Cosmotechnic Fusion, Justice AI GPT, AI labor, cognitive extraction, algorithmic bias, structural racism, colonial infrastructure, epistemic justice