We are part of the records we keep
– Gayatri Spivak
Plural Knowledges, Plural Worlds
In Spanish, the word for “knowledge” can also appear in the plural—conocimientos, “knowledges.” And there is another term, saberes, that conveys an even deeper dimension of knowledge, encompassing multiple forms of wisdom. When I type “knowledges” in English, autocorrect marks it as an error. This small contrast shows how each language opens a window into a different world, and how these worlds—each offering its own way of understanding reality—are meant to coexist. Through that coexistence, they widen our possibilities for surviving as human beings. Put differently, diversity expressed through the plurality of cultures, races, genders, languages, knowledges, and forms of existence is essential to humanity’s survival.
Epistemicide—the killing, silencing, annihilation, or devaluing of a knowledge system—occurs when epistemic injustices become persistent and systematic, operating collectively as a structured oppression of particular ways of knowing. At a moment when we’re witnessing an intentional and intensified effort to erase difference, it’s crucial to recognize that the erasure of diversity, languages, and knowledges is also the erasure of people—and, ultimately, of the planet.
At a moment when we’re witnessing an intentional and intensified effort to erase difference, it’s crucial to recognize that the erasure of diversity, languages, and knowledges is also the erasure of people—and, ultimately, of the planet.
The concept of epistemicide helps illuminate this problem. Epistemicide, “the death of the knowledge of the subordinated culture, and hence, the death of the social groups that possess it”, is directly connected to genocide and, in turn, to ecocide. Today, the war waged by the powerful against Otherness is a war increasingly amplified by artificial intelligence.
Cognitive Territories, Translation, and Large Language Models
Translation exposes the incommensurable nature of languages. Languages aren’t equivalent, and the world each one captures reflects a distinct system of knowledge. As Yásnaya Aguilar—Mixe linguist, writer, and activist—reminds us, language is a cognitive territory. So when we talk about translation, we’re talking about carrying one world into another, and—by definition—a complete translation is impossible, because our cognitive worlds cannot be fully aligned.
I want to begin with this reflection on language and knowledge because, when we talk about capitalism’s newest toy—AI—through chatbots and Large Language Models, we often assume that these systems somehow contain the world. But we know a model is never the world itself. A model is a representation of reality, but only of the slice of reality it is trained on and capable of expressing.
A model is a representation of reality, but only of the slice of reality it is trained on and capable of expressing.
When we talk about AI’s flattening of knowledge, the proposed solutions tend to be technical: making datasets more diverse, adding more Indigenous languages, or tuning the models to be less exclusionary of difference. What I don’t hear is a conversation about what it actually means to homogenize humanity’s knowledge—how that homogenization happens, and what consequences it carries.
The Three Traps of AI
With this in mind, I want to address three traps that I consider important to examine: universality, monoculture, and openness.
The first trap: Universality
The term “universal” derives from the Latin universalis, meaning “of or belonging to all,” which itself comes from universus, meaning “all together” or “whole.” However, as Western empires emerged through processes of colonization, the notion of universality became aligned with the worldview of the colonizer. The colonizer’s knowledge was positioned as superior, while anything that diverged from it was categorized as inferior—an assessment that applied both to the knowledge of the colonized and to the people who embodied that knowledge. Within the context of empire, this conception of the human as the center of the universe fosters the belief that difference must be erased, destroyed, or eliminated. As a result, we remain confined within a particular understanding of the universal, one that overlooks how the very idea of universality was appropriated and shaped by Western culture.
The second trap: Monoculture
We often think about culture in simplified terms. The word comes from the Latin colere, “to cultivate,” initially referring to cultivating the soil to raise a seed, and later used to describe the outcomes of social interaction: the processes, practices, beliefs, and creations that emerge from collective life. However, when we refer to culture, we often separate it from its materiality, its historicity, and the social and material conditions that make its production possible. As a result, culture becomes detached from us and turns into a commodity, and discussions about culture often implicitly reference a singular, dominant cultural way of relating to the world. The consequences resemble those of monoculture in agriculture, which we know to be ecologically and socially detrimental. When technological systems are treated as expressions of a hegemonic culture, they reproduce and reinforce technological monocultures, narrowing the diversity of ways in which humans understand and inhabit the world and homogenizing the ways humans live, feel, know, and exist.
The third Trap: Openness
Societies were born open. Cultures were born open. Capitalism, however, required the closure of culture and knowledge, because closed knowledge becomes a source of value accumulation. We often say “knowledge is power” or “information is power,” and both statements hold true, since knowledge and information can be readily transformed into resources, relationships, influence, or decision-making advantages that place some individuals above others. This is precisely why knowledge had to be restricted and accessible only to a few. Many years ago, we were among those who believed that, even though capitalism is a system that absorbs and consumes everything, it was necessary to fight for open knowledge and free culture. Openness seemed essential for ensuring that our knowledges could become visible and recognized in the world. Yet capitalism is a creature with many heads—what the Zapatistas (2016) call the capitalist hydra—and it appropriated openness as well. Now, the open is the most desired characteristic of the world. If the world is open, if we are open, if our data is open — anything is possible for AI
So, the question is: What do we do?
Pluriversality against Epistemicide, Genocide, and Ecocide
We often frame the problem as one of business models. The model is fundamentally extractive—we know this—and it absorbs every dimension of social, human, and natural life. The common assumption is that if we change the business model, the problem will be resolved. However, I am not convinced that changing the AI business model will fundamentally alter this reality, because the core issue is how Western racial supremacy, through its colonial mechanisms, continually develops new ways to optimize capitalism and its forms of dispossession, control, and violence over time.
For us, the problem—of universal knowledge presented as privileged knowledge imposed on the rest of us, together with techno-monocultures and forms of openness designed for AI capture and extraction—raises the question of which knowledge will be privileged. It pushes us to consider whose knowledge will be captured, registered, and regenerated by AI systems, which means that some knowledge will be shared, recorded, and preserved, while other knowledge will be erased, made invisible, or forgotten, along with the people that produced it.
What will happen to the knowledge of those who cannot speak or cannot live, those who are unable to own or govern the record of their memories and experiences?
What will happen to the knowledge of those who cannot speak or cannot live, those who are unable to own or govern the record of their memories and experiences? What if, one day, all our digital archives disappear? And if we do not preserve knowledge, we will not be able to preserve the people themselves, nor the ways they experience the world or care for their environments in order to survive. In the end, what I want to emphasize is that knowing cannot be separated from being, feeling, and perceiving the world. If we fail to recognize that knowledge is fundamental not only to the preservation of humanity but also to the preservation of the planet, we risk condemning ourselves to disappear from the face of the Earth.
This text was originally presented as a lightning talk at the conference ‘Towards Regenerative AI: Frames for Inclusive, Indigenous, and Intentional Innovation’ organized by IT for Change in Bengaluru on November 1, 2025. It has been revised for publication.