Beyond the AI Bubble: Why the Global South Needs a Human-Centered Approach Rather Than a Technological One

Over the past few years, artificial intelligence (AI) has evolved from a technological development into a global discourse presented as the next transformative revolution capable of reshaping economies, labor markets, states, and societies. Since the emergence of Large Language Models (LLMs) and the rapid expansion of generative AI applications, expectations surrounding the imminent arrival of Artificial General Intelligence (AGI) have intensified, accompanied by widespread predictions about the replacement of human labor across multiple sectors.

Yet, beneath the exaggerated optimism surrounding AI lies a deeper structural crisis related to the limitations of current AI models, the failure of many practical applications, and the enormous economic and energy costs that increasingly threaten the sustainability of the existing technological paradigm itself.

Within this context, a more sensitive dilemma emerges for the Global South, where many developing countries have begun adopting national AI strategies under the pressure of global competition and fears of technological marginalization. However, this accelerated embrace of AI risks reproducing new forms of technological and economic dependency, particularly if AI is treated not as a tool to support human development, but as a substitute for it.

Accordingly, the central debate is no longer confined to the capabilities of AI itself, but rather concerns the broader developmental model that countries of the Global South seek to pursue: will it remain centered on human development and social inclusion, or will it increasingly become shaped by the priorities of major technology corporations and the logic of the global digital marketplace?

First: The Crisis Between Technological Discourse and Practical Reality

Despite the optimistic rhetoric promoted by major technology companies, practical experiences increasingly suggest that AI applications continue to face profound structural limitations. Many promises concerning automation and the replacement of human labor have failed to materialize at the scale initially predicted.

Numerous studies have shown that the use of AI tools in fields such as software development and digital services does not necessarily improve efficiency; in some cases, it has actually slowed productivity and increased errors. Moreover, a considerable number of AI projects implemented within corporations have either failed or been scaled back after practical experimentation, particularly in sectors requiring substantial human interaction, ethical judgment, or professional sensitivity.

The core problem lies in the fact that LLMs do not “understand” the world in a human sense. Rather, they rely primarily on statistical prediction of linguistic patterns. Consequently, their ability to generate convincing responses does not equate to genuine reasoning, comprehension, or independent analytical capacity. This explains the phenomenon commonly referred to as “AI hallucinations,” in which systems generate false or misleading information in highly persuasive forms.

The risks become significantly greater when AI systems are deployed in critical sectors such as healthcare, education, and public administration. In these contexts, errors are no longer merely technical flaws; they may directly threaten people’s lives, rights, and economic opportunities.

Second: The Political Economy of the AI Bubble

The current AI boom cannot be understood independently from the political economy of technology. Major corporations operating in the AI sector are not promoting AI merely as a scientific breakthrough, but as an instrument for restructuring global markets, expanding economic influence, and consolidating control over data and digital infrastructure.

This race has contributed to the emergence of what may be described as an “AI bubble,” characterized by massive investments in data centers and large-scale AI infrastructure despite the limited financial returns relative to the scale of expenditures. Many AI companies continue to depend heavily on speculative investment and continuous financial expansion rather than on sustainable economic models.

At the same time, the rapid expansion of AI infrastructure raises mounting environmental concerns. Large data centers consume enormous amounts of electricity and water while contributing to carbon emissions and environmental degradation. As a result, the model of unlimited AI hyperscaling is increasingly being questioned from both economic and ecological perspectives.

Ironically, many developing countries—already struggling with energy shortages and infrastructure deficits—may find themselves pressured to bear the costs of digital transformation without possessing the structural capacity necessary to compete within the global technological economy.

Third: Why AI Fails to Replicate Human Intelligence

The limitations of current AI systems stem fundamentally from their underlying architecture. LLMs operate primarily through probabilistic pattern recognition across massive datasets, yet they lack consciousness, contextual understanding, lived experience, and genuine human cognition.

Accordingly, what often appears to be “thinking” is in reality a highly advanced process of data recombination rather than authentic reasoning. This explains why these systems continue to struggle with novel, complex, or contextually layered situations.

Furthermore, strong performance in benchmark evaluations does not necessarily indicate genuine intelligence. Much of the apparent sophistication of AI systems derives from advanced memorization and pattern retrieval capabilities rather than independent logical reasoning.

For this reason, skepticism is growing within scientific circles regarding the possibility of achieving AGI through the mere expansion of computational power and data scaling. Increasingly, researchers argue that current generative AI models may have already reached significant structural limits, and that future breakthroughs may require entirely different cognitive architectures beyond the present paradigm of generative AI.

Fourth: The Global South Between Digital Dependency and Human Development

The danger of the current global AI narrative lies in the pressure it places on developing countries to rapidly adopt AI technologies under fears of “falling behind.” However, this approach risks deepening new forms of digital dependency.

Control over advanced AI models and large-scale computational infrastructure remains concentrated in the hands of a small number of powerful corporations and technologically advanced states. Consequently, developing countries risk becoming passive consumers of AI technologies rather than active participants in their production and development.

Moreover, the uncritical expansion of automation within developing economies may intensify unemployment and social vulnerability, particularly in labor-intensive economies heavily dependent on low-skilled service sectors.

Therefore, there is an urgent need for a different AI model within the Global South—one that prioritizes the use of technology to support human development rather than replace it. Such a model would focus on low-risk, high-benefit applications, including digital translation, educational support systems, and scientific research assistance, rather than pursuing large-scale automation with potentially severe social consequences.

Fifth: Toward a Human-Centered Approach to AI

The real challenge facing the Global South does not lie merely in acquiring AI tools, but in the ability to integrate them into a developmental framework that preserves the centrality of human beings.

AI can undoubtedly serve as a powerful instrument for supporting scientific innovation, improving services, and enhancing administrative efficiency. However, it cannot function as a substitute for human agency, social policy, or equitable development strategies.

National AI strategies must therefore be linked to the protection of digital rights, the promotion of social justice, the preservation of employment opportunities, and the prevention of excessive concentration of data and technological power in the hands of transnational corporations.

In this regard, countries of the Global South may need to fundamentally redefine their relationship with technology. Development should not be measured solely by the speed of AI adoption, but rather by societies’ ability to employ technology in ways that genuinely serve human welfare and inclusive development.

Conclusion

The current AI wave reveals a central paradox: while AI is presented as the future of humanity, mounting evidence suggests that the existing model suffers from deep technical, economic, and structural limitations.

For the Global South, the primary danger lies not only in technological backwardness, but also in uncritical integration into a digital model that may reinforce dependency and social fragility rather than alleviate them.

Consequently, the challenge is not to reject AI altogether, but to redirect it within a more balanced human-centered developmental vision—one that places human beings at the center of development rather than reducing them to mere variables within algorithmic systems.