Vishal Reddy is a second-year master’s student at the Department of Geopolitics and International Relations, Manipal Institute of Social Sciences, Humanities and Arts, Manipal Academy of Higher Education (Institution of Eminence), Manipal, India.
Contemporary Artificial Intelligence (AI) development increasingly extends beyond isolated technological innovation, emerging instead through complex interactions among private firms, computational infrastructures, data systems, algorithms, organizational structures, and human expertise integrated within broader technological ecosystems. These interactions highlight the increasingly networked character of AI development, where technological progress depends upon relationships among multiple interconnected actors and infrastructures.
Within these evolving ecosystems, private sector companies increasingly function as the primary engines of AI development due to their massive infrastructural, financial, computational, and organizational capabilities. This aligns with the argument of Cronin, who emphasizes that private sector companies increasingly operate as geopolitical actors because of their growing control over infrastructures and domains of society such as communication networks, digital infrastructures, commerce, and informational systems that were historically associated with the authority of states. In the context of AI development, this centrality is further reinforced through the financial capabilities of Google, Amazon, Facebook (Meta), Apple, Microsoft (GAFAM) firms toward large-scale AI infrastructure expansion, positioning them at the core of contemporary AI ecosystems.
From the lens of Bruno Latour’s Actor-Network Theory (ANT), these developments reflect how AI systems emerge not through isolated technological innovation alone, but through interactions among heterogeneous human and non-human actors embedded within broader socio-technical networks. This further aligns with the ANT concept of “punctualization”, which refers to the process through which complex technological systems are represented as singular and homogeneous entities while concealing the heterogeneous interactions and structures embedded within them. Viewed through the lens of punctualization, technologies such as AI are often projected by firms, policymakers, and public narratives as unified and autonomous systems, hiding the underlying infrastructures, human labor, organizational hierarchies, computational systems, datasets, and financial structures that sustain them. However, moments of disruption within these organized technological networks de-punctualize the system and expose the complex socio-technical nodes operating beneath the appearance of technological coherence.
This dynamic is particularly evident in the case of Meta Platforms, especially following the company’s recent massive layoffs of approximately 8,000 employees. Alongside this development, there was a leaked internal recording of Mark Zuckerberg, Chairman and CEO of Meta, discussing the role of employee activity and internal organizational knowledge in training AI models. In the recording, Zuckerberg emphasized that AI models “learn from watching really smart people do things”, further arguing that the concentration of highly skilled engineers within Meta enables the company to accelerate the coding capabilities of its AI systems more effectively than competitors lacking similar organizational depth and technical talent. These remarks reveal the underlying socio-technical architecture embedded within AI systems, demonstrating that AI development does not emerge autonomously through algorithms alone, but rather through interactions among human expertise, behavioral monitoring systems, computational infrastructures, and large-scale financial investments.
In this context, the leaked recording and the subsequent layoffs function as a process of de-punctualization, exposing the concealed complexities under the apparently seamless representation of AI systems. While AI models are publicly projected as intelligent technological entities, the Meta case reveals the dependence of these systems upon extensive human participation, monitored data extraction, engineering labor, organizational coordination, and infrastructural expansion. The ethical implications of monitoring employee behavior and internal work processes to train AI systems further demonstrate how human activity itself increasingly becomes integrated into AI development networks. Simultaneously, the deployment of AI systems contribute towards restructuring labor relations through workforce reductions, reflecting the contradictory dynamics integrated within contemporary AI-driven growth – where human labor remains essential for training and optimizing AI systems, even as these same systems increasingly contribute toward labor displacement and organizational restructuring.
The case also reinforces the infrastructural dimension of contemporary AI development. In the leaked recording, Mark identified the quantity and quality of compute, efficiency in compute utilization, and access to data as the core elements determining the strength of AI models. This directly aligns with the broader argument that AI competitiveness increasingly depends upon infrastructural and financial capabilities, rather than solely algorithmic innovation. Data and algorithms operate as the foundational components that enhance the efficiency and performance of AI systems, while the ability to expand hyperscale data centers, sustain cloud infrastructures, and recruit elite engineering talent increasingly determines competitive advantage within AI development. Consequently, AI power is becoming progressively concentrated within a small number of private firms possessing the infrastructural, computational, and financial capacity necessary to sustain large-scale AI ecosystems.
Thus, viewed through the lens of ANT, contemporary AI development cannot be understood purely through conventional state-centric or technologically deterministic explanations. Rather, AI development emerges through interactions among private firms, engineers, infrastructures, computational systems, algorithms, datasets, organizational networks, and human actors embedded within broader socio-technical systems. Within these evolving networks, private technology firms increasingly occupy a central position due to their control over the infrastructures, computational resources, financial capabilities, and organizational systems essential for contemporary AI development.
Disclaimer: The views expressed in the article are personal.
