Artificial intelligence (AI), because of its structural dependence on US English, initiates a new era of domination on language distinct from historical colonialism. AI poses a new, powerful threat of linguistic, technological, historical, ideological, and cultural homogenization. While AI and colonialism share a power dynamic of dominance and extraction, AI contradicts and suppresses a hybridity that was characteristic of relationships between colonized and colonizer in previous centuries. The paper thus examines the colonial, postcolonial, and AI theorists' views on colonial power, hybridity, culture, and hegemony, and argues that postcolonial theories alone cannot fully capture AI’s singularity—a one‑directional, non‑negotiated interaction.

This shift becomes especially evident in the context of Bangladeshi English (BdE), which explains how AI dominance can homogenize language and culture in South Asia and across the Global South. In the course of more than two hundred and fifty years, the people of what is now Bangladesh have established is a balanced, decolonial relationship between English, the language of the most recent colonizer, but also an access point to global commerce, and Bangla, the language of the majority of inhabitants in this part of South Asia — a situation perhaps most memorably described by Nobel Prize-Winning Bangla author Rabindranath Tagore as the intersection of the country’s two great rivers, the Ganges and the Jamuna. While Tagore’s metaphor captures the interactive hybridity of colonial linguistic exchange, AI resembles something very different, operating more like the Mississippi. AI is built on algorithmic frameworks that absorb data globally and homogenize it into a single linguistic variety, as the Mississippi carries everything toward a single dominant stream

To address theoretical gaps in postcolonial theory for understanding the new hegemony of AI, this paper presents two conceptual tools. The first, Kulturraum, a geographical and social space with a shared set of cultural practices and values, reveals how hybridity creates an established mixed culture, such as Bangladeshi English (BdE). The second—the Company Town, which refers to a corporate entity that dominates a community—explains the convergent, non-negotiable logic of corporate technology design. These two concepts demonstrate that though AI technology reproduces colonial patterns, it erases the established hybrid culture that emerged from colonial interaction. The paper concludes that this non-interactive singularity of AI threatens to extinguish hybridity and poses an existential risk to the cultural dynamism of postcolonial societies like Bangladesh.

This paper argues for a decolonial digital humanities that treats computational "failures" as pedagogical opportunities rather than problems to solve. Drawing on recent topic modeling research of al-Risālah, a foundational magazine of the Arab Nahḍah, this research demonstrates how traditional digital humanities approaches reproduce orientalist frameworks by forcing non-Western texts into Western epistemological categories. When algorithms struggle to process the magazine’s sophisticated synthesis of Islamic, Arab, and European intellectual traditions, their breakdowns reveal the violence of colonial classification systems.

I propose three methodological interventions: critical visualization that foregrounds opacity rather than transparency; treating algorithmic silences as data about alternative knowledge systems; and using computational reading to unsettle rather than confirm inherited narratives. Through analysis of Ahmad Hasan al-Zayyāt's editorial practice in al-Risālah, I show how his "networked self," constituted through relationships with communities and traditions rather than individual authorship, challenges the possessive individualism embedded in most digital humanities tools. The paper concludes by outlining pedagogical approaches that teach students to read digital archives against their grain, arguing that the future of digital humanities lies not in making Global South texts more accessible to Western analysis, but in allowing these texts to transform our analytical frameworks themselves.

As the National Education Policy (NEP) 2020 mandates the integration of Indian Knowledge Systems (IKS) into mainstream curricula, educational administrators are increasingly turning to digital platforms like SWAYAM and DIKSHA to scale this dissemination. However, this digitization process represents a profound "translation"—not merely between languages (e.g., Sanskrit to English), but between epistemologies: from the embodied, oral tradition of Guru-Shishya Parampara to the standardized, modular format of the Learning Management System (LMS). Drawing on Barbara Cassin’s concept of the "untranslatable" as that which one "never stops (not) translating," this paper investigates the pedagogical "residue" lost when India’s cultural heritage is remediated by algorithmic logic and Large Language Models (LLMs).

Thesis

This paper argues that the "untranslatable" in the context of Indian Digital Humanities is the oral and situational context of learning. While LLMs and digital platforms can successfully translate the informational content of IKS (texts, dates, formulas), they systematically fail to translate the pedagogical process—specifically the dialectical and non-linear methods inherent to these traditions. This failure is not a technical glitch but a fundamental clash between the "probabilistic" nature of AI (which seeks standardized averages) and the "singular" nature of the Guru (teacher-student specificity).

Methodology

The research employs a comparative digital ethnography focusing on the "Introduction to Indian Philosophy" and related IKS courses currently hosted on the SWAYAM/NPTEL platform. The methodology is threefold:

1) Curricular Analysis: Examining how linear LMS structures (modules, quizzes) force cyclical IKS concepts into linear, assessment-driven frameworks.

2) Algorithmic Auditing: Subjecting specific IKS concepts (e.g., Pramana or Dharma) to leading LLMs (GPT-4, Claude) to identify "hallucinations of simplifications" -instances where the model flattens complex philosophical debates into binary Western equivalents to satisfy the prompt.

3) Policy Review: Analyzing the gap between the NEP 2020’s vision of "holistic education" and the administrative reality of "data-driven" digital pedagogy.

Conclusions & Implications

Preliminary analysis suggests that the current digital administration of IKS prioritizes "access" over "authenticity," resulting in a "zombified" digital heritage that looks preserved but is pedagogically inert. The paper concludes by identifying a critical innovation gap: the need for "Sovereign Digital Architectures." These are proposed digital pedagogical tools designed specifically for the Indian context that resist total translation—platforms that preserve the "friction" of learning by refusing to simplify complex cultural concepts for the sake of user experience (UX). The stake of translation here is the preservation of cognitive diversity; if we render IKS entirely "translatable" to AI, we risk colonizing the very tradition we aim to preserve.