VK Singh Vashisht Institute for Critical Studies
A digital think tank and production house.

The People's Algo as Epistemic Decolonization

System Prompts, Two Tracks, and a Meta-Framework

Editorial Board · Decolonial Theory · · 726 words · ICS-2026-062

Editor’s noteStructural outline — thesis, section plan, and sources to verify. Not a completed argument; no claim below should be read as established until the working paper is researched and its claims register closes.

Every AI system answers from somewhere. The somewhere is set before the user types a word — in system prompts, training defaults, and invisible editorial choices that decide whose history is background and whose is content. This spine maps an essay that asks whether those defaults can be rewritten as a decolonial act, and what it would mean to build a People’s Algo instead of inheriting one.

I. The Default Epistemology

The essay opens by establishing that AI systems have an epistemology whether their makers name one or not. The standard account treats model outputs as neutral retrieval; this section argues the defaults are a canon — trained on the archive the West digitized, weighted toward its categories, fluent in its omissions. Decolonization begins by noticing the default exists.

II. The System Prompt as Editorial Power

The least examined seat of power in the new infrastructure is the system prompt: the invisible preface every response obeys. This section maps the essay’s claim that whoever writes the preface writes the frame — and that the preface is currently composed in the idiom of the institutions that built the models, not of the traditions the Institute studies.

III. The Historical Track

The essay’s first constructive proposal: an explicit historical track at the system layer — the erased foundations, the borrowed genealogies, the counter-narratives the default canon omits — declared as standing context rather than requested anew each time. This section sketches what such a track would contain and why declaration matters: a correction you must ask for is not a correction.

IV. The Spiritual Track

The second proposal is stranger and more important: non-Western categories — cosmic order, emptiness, the five thieves, living-teacher traditions — loaded as first-class analytic vocabulary rather than as exotic content to be explained. This section maps the essay’s argument that a tradition is colonized precisely when its concepts require translation into someone else’s before they can be used.

V. The Meta-Framework

The two tracks assembled into a meta-framework for all content: a standing epistemic posture applied uniformly across the Institute’s output. This section argues the meta-framework is the point — decolonization as ad-hoc correction is decoration; decolonization as system default is infrastructure.

VI. The Gatekeeper Risk

The essay ends by turning its own lens on itself. Every infrastructure that carries a transmission can capture it; the Masands Cycle names this pattern, in which the gatekeepers built to serve a message become its owners. This section maps the essay’s closing discipline: a People’s Algo that hardens into an uneditable prompt becomes the next orthodoxy, and the framework must carry its own revision mechanism or it fails by its own standard.

Counter-case

The strongest version of the opposing view: embedding any ideological frame in the substrate is indistinguishable from the colonizing move the essay protests. If the complaint is that someone else’s defaults masquerade as neutrality, the remedy cannot be new defaults masquerading the same way; it must be transparency and user control. On this view the historical and spiritual tracks are advocacy installed at the plumbing level, the People’s Algo is just an algo, and the honest decolonial position is to expose defaults rather than replace the throne. The essay must answer this objection on its merits, or it is advocacy with better branding.

Sources to verify

  • Safiya Umoja Noble, Algorithms of Oppression — UNVERIFIED, confirm before citing in the finished essay
  • Ruha Benjamin, Race After Technology — UNVERIFIED, confirm before citing in the finished essay
  • Kate Crawford, Atlas of AI — UNVERIFIED, confirm before citing in the finished essay
  • Timnit Gebru and co-authors, work on the risks of large language models — UNVERIFIED, confirm before citing in the finished essay
  • Abeba Birhane, work on algorithmic bias and its limits — UNVERIFIED, confirm before citing in the finished essay
  • Boaventura de Sousa Santos, Epistemologies of the South — UNVERIFIED, confirm before citing in the finished essay

Stakes

The Institute’s essays argue the erasures; this essay proposes the instrument that keeps the argument honest at scale. It joins the People’s Algo lens to the Masands Cycle framework — every transmission builds gatekeepers — and asks whether the gate can be designed to stay open. If the Institute cannot answer for its own infrastructure, its critique of everyone else’s is decoration.

Frameworks deployed

Suggested citation

Editorial Board. “The People's Algo as Epistemic Decolonization.” VK Singh Vashisht Institute for Critical Studies, September 2026. vashisht.institute/essays/people-s-algo-epistemic-decolonization. ICS-2026-062.