Generative AI at the Crossroads: From the North's Old Certainties to the South's New Possibilities

What This Text Presents:

A critical analysis of how generative artificial intelligence (GAI) reproduces patterns of digital colonialism, based on:

1. Exploitation of the Global South

Data extraction (treated as free raw material)
Mineral extraction (e.g., cobalt from the Congo)
Precarious labor in AI training (e.g., underpaid annotators in Kenya or India)

2. Alternatives from the South

China (DeepSeek – efficient AI using local data)
India (Bhashini – AI for marginalized languages)
Africa (Lelapa AI – community-driven AI development)

3. Keys to Ethical AI

Technological sovereignty (reducing dependence on the Global North)
Situated ethics (integrating local knowledge, not just "universal solutions")
Sustainable models (lower energy use, reduced extraction)

Conclusion: AI is not neutral. While the Global South is already building alternatives, the challenge is avoiding the same domination patterns.

Why This Matters:

Reveals AI’s geopolitical impacts (it’s not just "technology")
Proves other AI models are possible (beyond Silicon Valley’s approach)
Inspires public policies and tech development in Latin America and other Global South regions.

Key Terms:

Digital colonialism = Tech dominance replicating historical exploitation
Situated ethics = Context-aware, culturally grounded AI design
Technological sovereignty = Self-determined tech development

In his lecture "AI in a Post-Reason Era", Ben Vinson III, President of Howard University, argued that artificial intelligence must be "developed with wisdom," with the central message that technological progress should serve humanity. Citing the philosopher Cicero and his idea that the good life is centered on the pursuit of virtue and wisdom, Vinson questioned whether AI could truly be used for humanity's benefit or whether, conversely, it might represent a threat to critical reflection and human agency.

In his MIT Compton Lecture, he analyzed the development of generative artificial intelligence (GAI) from a historical-critical perspective, highlighting its sociopolitical implications. As president of Howard University, Vinson examined how GAI could reproduce structural inequalities if its development is divorced from a historical analysis of knowledge and power systems. His approach situates the discussion in the context of North-South relations, questioning the asymmetric distribution of influence over technological direction.

The "Age of Reason" and Its Paradoxes

Vinson problematized Western rationalism of the 17th and 18th centuries, a period that, while celebrating reason as a liberating force, restricted this concept of freedom to a few. This exclusion was not accidental: it served to justify atrocities such as slavery and colonialism, under the pseudoscientific argument that certain racial groups were "incapable of full rationality." This colonial discourse, for example, replaced initial questioning about whether colonized peoples had souls with "scientific" questioning about reasoning capacity, determined by fictitious biological and environmental criteria - a logic that echoes today in algorithms that classify and predestine individuals. By reducing human beings to hierarchical categories, this thinking not only legitimized domination but also naturalized the dehumanization of entire populations. And this history dates back to the birth of the scientific method, preceded by the invention of a rather peculiar concept of nature.

#Francis Bacon and the Birth of the Modern Concept of Nature

At the genesis of this history is Francis Bacon (1561-1626), who, in addition to being an English philosopher and statesman, also served as Attorney General and Lord Chancellor, acting as administrator of justice. As a royal landowner, his duties included overseeing judicial practices, which involved monitoring interrogations and torture used to extract confessions. This part of his biography was more influential than is usually acknowledged regarding his version of the experimental method, of which he is often considered one of the creators. In works such as Novum Organum (1620), Bacon advocated a new approach to understanding and controlling nature. Believing that "knowledge is power," he constructed arguments such as science's purpose being to "conquer nature" for humanity's benefit. His vision was rooted in the idea that nature was a resource to be dominated, subordinated to human will.

Bacon's philosophy marked a significant shift from earlier, more relational views of nature found in indigenous, Eastern or even medieval European cosmologies. Instead of seeing nature as a living, interconnected system, Bacon framed it as a passive object to be dissected, analyzed and exploited. This mechanistic view of nature set the stage for the Scientific Revolution and Enlightenment, which further consolidated the idea of human dominion over the planet.

The Modern Notion of Nature and Its Historical Context

The modern notion of nature emerged in the 16th and 17th centuries, a period marked by profound social, economic and intellectual transformations. This era saw the rise of capitalism, the expansion of European colonial empires, and the consolidation of the nation-state. The fragmentation and hierarchization of human-nonhuman relations were not mere philosophical abstractions but were deeply intertwined with these historical processes.

The capitalist mode of production, beginning to take shape during this period, depended on the extraction and commodification of natural resources. Nature was reconceptualized as a reservoir of raw materials destined to fuel industrial growth and economic expansion. Bacon's vision of "conquering nature" aligned perfectly with this capitalist ethos, providing philosophical justification for unlimited exploitation of the natural world (which was, after all, an infinite resource, wasn't it?).

Colonialism and the Domination of the "Other"

The modern notion of nature was also deeply connected to the colonial project. European colonial powers saw the lands and peoples they encountered as part of a "wild" and "uncivilized" nature that needed to be "tamed" and "controlled." Indigenous knowledge systems, which often emphasized the interconnection between humans and non-humans - an obvious recognition that we are part of the same nature - were dismissed as primitive or superstitious. This hierarchical view of nature extended to the dehumanization of colonized peoples, who were often treated as part of the natural world to be dominated and exploited.

With the advent of the scientific method, inspired in part by Bacon's ideas, the separation between humans and nature was reinforced. Thinkers like René Descartes further developed this dualism, framing nature as a machine governed by fixed laws. This mechanistic view not only justified the exploitation of nature but also facilitated the development of technologies that enabled colonial expansion, such as navigation tools, firearms, agricultural innovations and social control techniques.

Generative AI: More Continuity Than Rupture

Fast forward to the GAI boom of November 2022, we see that Generative Artificial Intelligence does not emerge in a vacuum, but rather from material and human conditions deeply rooted in global relations of exploitation, based on a capitalist mode of production that was born "written in the annals of humanity in letters of blood and fire."

We observe then that GAI has as its conditions of possibility the development of increasingly powerful hardware (dependent for its production on the intensive exploitation of minerals) and intensive human labor, both indirectly - through intellectual production abundantly available online - and directly - through poorly paid and unhealthy human labor for AI training.

Regarding the intensive extraction of minerals for hardware, this process frequently occurs in the Global South under precarious environmental and social conditions, generating disproportionate impacts in the regions where it takes place, while also creating a cycle of hardware production and consumption that causes even greater harm to the Global South, since, as studies indicate, the electronic waste resulting from this production chain is mostly sent to developing countries, where its "recycling" exposes workers to toxic substances, completing a cycle of extraction, production concentrated in the Global North, and disposal in an unequal value chain.

Beyond material exploitation, AI feeds on millions of online intellectual contributions, which are transformed into "free raw material" to train algorithms. Even more serious is the precarious work of data annotation, performed by workers in often unhealthy conditions, with low wages and no labor protections - an activity essential for the functioning of AI systems that are later commercialized by Global North corporations.

This dual exploitation, of both natural resources and labor, reveals a map of geopolitical asymmetries embedded in contemporary technological development.

Thus, the current AI paradigm is founded on unequal relations that reproduce historical dynamics of domination. While economic benefits and technological control remain concentrated in the Global North, human and environmental costs fall disproportionately on the Global South. Recognizing these material conditions is fundamental for any project seeking truly ethical and just AI that overcomes the extractivist and colonial logic still present at the heart of the digital revolution. This seems to be a "blind spot" in Vinson's discourse - he fails to recognize that building alternatives will require not just technical or even ethical changes, but above all the transformation of these structural power relations.

Between Heidegger, Deleuze & Guattari and Álvaro Vieira Pinto: The Problem of Technology

Although his critique of generative AI touches on the risks of a dehumanizing rationality, Vinson's speech does not constitute a rejection of technology, but rather an ethical call for its reinvention. At this point, we draw on some history and the contrast between thinkers Martin Heidegger on one side, and Gilles Deleuze, Félix Guattari and Álvaro Vieira Pinto on the other, to understand possible paths forward.

Heidegger, an important Western thinker on technology, approached it as "enframing" (Gestell), warning of the risk of reducing the world to mere resource. For him, modern technology conceals other ways of revealing being, trapping thought in utilitarian logic. Deleuze and Guattari, in contrast, see in technology and machines not just instruments of domination, but also possibilities for creation, multiplicity and escape. The assemblage is a connection of heterogeneous bodies, forming a network of forces through proximity, sympathy or symbiosis. It is the minimal unit of enunciation production, operating simultaneously as a machine of bodies and a collective of enunciation. These two dimensions are distinct, autonomous, but inseparable, articulated through conjugation or neighborhood. Being thus a cutout of a network of relations, it seems to provide broad possibilities for readings and propositions regarding the variability of aspects or dimensions of reality in their relations, which is convenient for a use of technology that is not limited to reproducing structures, but that invents new ways of existing.

Meanwhile, Álvaro Vieira Pinto offers an even more affirmative perspective. Developing the concept of "amanualidade", he highlights the active role of individuals in technological creation and forms of social production. For him, technology is an expression of human knowledge embedded in concrete historical contexts - not an autonomous or alienating force. Pinto proposes critical reflection without technophobia: he denounces both technological fetishism and its total rejection, advocating for ethical, situated and conscious use. His critique of "technological dazzlement" resonates as a contemporary warning against uncritical belief in AI's miraculous powers.

Thus, Vinson's critique echoes Heidegger in denouncing the dangers of AI repeating the colonial project of total rationalization. But it should also align with Deleuze and Vieira Pinto in suggesting that technology can be reappropriated as a force of resistance, as a vector of difference and emancipation. Deleuze and Guattari in particular offer a key for thinking about distributed agency in technical processes: rather than asking whether AI has agency, they invite us to observe what emerges from the assemblage between humans and machines. In this interaction, affects emerge, new possibilities, new forms of creation - and it is this field of forces that interests us politically.

Global North, Global South: Two Projects for the Future

The true opposition emerging in discussions about generative AI is not between technology and humanism, but between two historical modes of producing and imagining the future. The dominant model, gestated in the Global North, is deeply rooted in a modern rationality that separates subject and object, human and nature, knowledge and power. This rationality - which shaped modern science, capitalism and colonialism - operates through a logic of domination and accumulation, where technology serves efficiency, control and the reproduction of inequalities.

On the other hand, the Global South - understood here not just as geographic location but as geopolitical and epistemic position - offers not a negation of technology but the possibility of another use, ethical and situated. In many Southern cultures and worldviews, technology is thought in relation to collectivity, to territory, to long-term temporalities. Technical production does not occur despite life, but in dialogue with it. In this sense, the Global South presents itself not as passive victim of others' progress, but as space for creating alternatives. An AI forged from these perspectives could, for example, incorporate values of reciprocity, care, ancestrality, "cacharreo" (tinkering) and environmental justice - directly challenging the extractivist model in place.

Vinson's critique, therefore, is not just about AI's risks, but about the possibility of refounding it from other possible worlds. The hope lies in recognizing that the Global South need not merely resist the North's technical project - it can create others, with other purposes, vocabularies and temporalities. Herein lies the deepest political potency of this moment we're living: to imagine technologies that serve life, not the other way around.

Material Signs of a New Technics: Toward Decolonized AI

The critiques presented here of the hegemonic model of artificial intelligence development - marked by data extractivism, unsustainable energy consumption and technopolitical centralization - are not limited to the theoretical realm. Concrete initiatives are emerging from the Global South that gradually materialize viable alternatives, demonstrating in practice that other ways of doing are possible. These cases reveal not just technical changes but a profound epistemological and political repositioning in the relationship between technology and society.

In China, the DeepSeek project represents a break with the dominant model. While systems like GPT-4 require massive computational infrastructure - with energy consumption comparable to small cities - DeepSeek was conceived from radically different principles. Its optimized algorithmic architecture reduces computational requirements by up to 70%, while its training process employs datasets filtered for cultural and linguistic relevance. This approach not only minimizes environmental impacts but avoids the indiscriminate appropriation of global content characteristic of the predominant extractivist model.

This is not an isolated initiative. China's AI ecosystem has been developing alternative paths that reject mere replication of Western models. Baidu's Ernie Bot, for example, was specifically designed to understand nuances of classical and contemporary Chinese, incorporating cultural values like the concept of hexie (social harmony) into its response structures.

These initiatives are supported by an innovative regulatory framework. The Generative AI Regulation (2023), for example, establishes that at least 30% of data used to train models developed in the country must be of domestic origin. This measure not only strengthens Chinese digital sovereignty but creates barriers against the neocolonial information extraction that characterizes many Western platforms.

The phenomenon repeats in other Global South regions. In India, the Bhashini project develops language models for 22 local languages, prioritizing rural speakers historically excluded from global digital ecosystems. In Kenya, startup Lelapa AI creates African language processing tools through collaborative fine-tuning processes that directly involve linguistic communities in technological development.

These examples share fundamental characteristics: (1) rejection of unlimited growth logic in favor of contextualized efficiency; (2) integration of local values and knowledge into technical development processes; and (3) establishment of governance mechanisms ensuring transparency and community control. The challenge is one of scale and sustainability. How to ensure these experiences don't remain isolated niches but can constitute alternative networks capable of challenging current technological hegemony? The answer will require not just technical advances but above all new forms of South-South cooperation, non-extractivist financing models and new and better regulatory frameworks.

We are not talking about choosing between different AI models - but rather deciding whether we'll be capable of building a truly plural and decolonized AI. The cases cited here demonstrate that, far from being mere repetitions of Global North innovation, Global South countries are leading some of the most interesting and promising experiments in technological reinvention. Perhaps what's happening today is concrete proof that other assemblages composed of humans and machines are already underway. In this sense, positioning myself as a South American researcher, I argue that it's time to understand what role our countries will play in this process.