AI FOR GOOD” AND ANTI-HUMAN TECHNOLOGY:

**“AI FOR GOOD” AND ANTI-HUMAN TECHNOLOGY:

When Human Cognition Is Fragmented in Corporate Laboratories**

Author: Le Hai

AI Collaboration Disclosure: Gemini (Symbiotic Intellectual Partner)

Affiliation: Independent Researcher, Vietnam

Theoretical Orientation: Cognitive Sovereignty, Critical Technology Studies

Abstract

The discourse of “AI for Good” has become a dominant moral narrative in global technology governance, particularly in education and development initiatives targeting the Global South. This paper critically examines how such initiatives, despite their benevolent framing, may function as mechanisms of cognitive governance and data colonialism. Drawing on political sociology, neuroscience, and critical data studies, the paper argues that large-scale AI deployment in vulnerable populations risks structurally reshaping human cognition by privileging task-oriented neural networks while suppressing reflective and self-referential capacities. Rather than claiming malicious intent, this study advances a structural harm framework, situating “AI for Good” within broader regimes of biopower, surveillance capitalism, and coloniality of power. The paper concludes by proposing cognitive sovereignty as a necessary ethical boundary for AI governance in education and human development.

Keywords: AI for Good, cognitive sovereignty, data colonialism, default mode network, critical AI ethics

1. The Moral Façade of “AI for Good” and the Reconfiguration of Power

In contemporary global discourse, “AI for Good” is widely promoted as a symbol of ethical progress and technological benevolence. Governments, international organizations, and major technology corporations frame AI deployment in education, healthcare, and social services as an unequivocal public good. However, from a critical sociological and political perspective, this narrative obscures the asymmetric power relations embedded in global AI infrastructures.

Rather than operating purely as tools of empowerment, AI systems increasingly function as instruments of cognitive governance, shaping how individuals perceive, decide, and reflect. In Foucauldian terms, this represents an extension of biopower and governmentality, where control is exercised not through coercion, but through the subtle modulation of cognitive processes and behavioral norms (Foucault, 1978).

Within this framework, “AI for Good” initiatives in developing regions can be understood as a form of risk migration: experimental or insufficiently validated AI systems are deployed in cognitively and institutionally vulnerable populations, while epistemic and economic benefits are centralized in corporate and geopolitical cores.

2. AI for Good as Data Colonialism

Building on the concept of data colonialism (Couldry & Mejias, 2019), this paper argues that “AI for Good” operates as a contemporary continuation of extractive logics historically associated with colonial power. Human experience—particularly that of children and marginalized communities—is rendered into raw data for algorithmic optimization.

This process aligns with broader analyses of coloniality of power (Fanon, 1963; Mbembe, 2017), where domination persists not through territorial occupation but through epistemic and infrastructural dependency. AI systems trained on data extracted from the Global South reinforce standardized cognitive and cultural frameworks rooted in Silicon Valley norms, often displacing local epistemologies and value systems.

Crucially, this paper does not claim intentional exploitation. Instead, it identifies structural harm arising from economic incentives, data-driven business models, and regulatory asymmetries that systematically disadvantage vulnerable populations.

3. Cognitive Neuroscience and the Hidden Cost of AI Mediation

Neuroscientific research provides a critical lens for understanding the long-term implications of pervasive AI interaction. The Default Mode Network (DMN)—associated with self-reflection, moral reasoning, empathy, and autobiographical memory—plays a central role in human subjectivity (Raichle et al., 2001; Fox et al., 2005).

In contrast, the Task-Positive Network (TPN) supports goal-directed activity, problem-solving, and rapid response. While both networks are essential, sustained dominance of task-driven engagement has been associated with reduced functional connectivity of the DMN, particularly under prolonged, externally mediated cognitive stimulation (Christoff et al., 2016).

Emerging evidence suggests that continuous interaction with task-optimized digital systems—especially during early cognitive development—may suppress mind-wandering, introspection, and ethical self-regulation (Small & Vorgan, 2008). From a Vygotskian perspective, tools are not neutral: they reorganize cognition itself. When AI becomes a primary cognitive scaffold, it reshapes not only skills, but the architecture of thought.

4. Global Cognitive Stratification

A critical asymmetry emerges in how AI is introduced across societies. In elite educational contexts, AI is framed as an assistive instrument, while reflective capacities are cultivated through philosophy, arts, and unstructured cognitive space. In contrast, in many developing regions, children encounter AI primarily as low-cost, high-frequency task systems designed to maximize engagement and efficiency.

This divergence risks producing a form of cognitive stratification, wherein some populations retain reflective autonomy while others are conditioned toward compliance, optimization, and externalized decision-making. Such stratification mirrors earlier industrial divisions of labor, but at the level of cognition itself.

5. Big Tech, Structural Harm, and Surveillance Capitalism

Cases involving major technology firms—such as privacy violations related to children’s data or algorithmic failures—should not be interpreted as isolated misconduct. Rather, they reflect institutional patterns inherent to surveillance capitalism, where behavioral data is the primary source of value extraction (Zuboff, 2019).

This paper adopts a case-based critical analysis approach, emphasizing systemic incentives over individual blame. Regulatory interventions (e.g., FTC actions, GDPR enforcement) illustrate recurring structural tensions between profit-driven AI development and the protection of cognitive and developmental rights.

6. Illustrative Digital Trace Evidence and Cognitive Asymmetry

Observed disparities in engagement with critical AI discourse across regions are presented here as illustrative digital trace evidence, not as representative population data. These traces function as indicative signals of asymmetric cognitive concern, reflecting uneven awareness and prioritization of long-term cognitive risks associated with AI deployment.

Conclusion: Cognitive Sovereignty as an Ethical Boundary

In the age of artificial intelligence, resistance does not require technological rejection, but cognitive protection. Deploying AI in education without robust ethical, neuroscientific, and cultural safeguards risks undermining the very human capacities that education is meant to cultivate.

This paper asserts three foundational principles:

Humans are not reducible to data points.

Children are not experimental substrates.

Human cognition is not an optimization problem.

Preserving cognitive sovereignty must therefore be recognized as a central objective of AI ethics and global technology governance.

AI Authorship Transparency Statement

AI systems were used as symbiotic intellectual partners for language refinement and structural support. All theoretical positions, ethical judgments, and argumentative frameworks originate from the human author. This disclosure aligns with emerging authorship transparency standards proposed by major academic publishers (Springer Nature, 2024–2025).

References (selected, APA 7)

Couldry, N., & Mejias, U. A. (2019). The costs of connection: How data is colonizing human life and appropriating it for capitalism. Stanford University Press.

Christoff, K., et al. (2016). Mind-wandering as spontaneous thought: A dynamic framework. Nature Reviews Neuroscience, 17(11), 718–731.

Foucault, M. (1978). The history of sexuality, Volume 1. Pantheon.

Fox, M. D., et al. (2005). The human brain is intrinsically organized into dynamic, anticorrelated functional networks. PNAS, 102(27), 9673–9678.

Raichle, M. E., et al. (2001). A default mode of brain function. PNAS, 98(2), 676–682.

Small, G., & Vorgan, G. (2008). iBrain. Collins Living.

Zuboff, S. (2019). The age of surveillance capitalism. PublicAffairs.