Gramsci was one of the most influential political theorists of the twentieth century. His ideas remain essential for understanding the relationship between power, culture, ideology and everyday life. This analysis is an attempt to understand how his concept of hegemony, the process by which dominant classes secure “active consent” for those they rule, may be applicable in a society dominated by artificial intelligence.
In his view, the power of the upper classes is not maintained solely through coercion or economic control. Rather, they achieve a stable dominance by shaping the cultural and ideological landscape in which people live. This produces a form of rule where morality, common sense, and social norms are gradually nudged towards people considering their own subjugation to upper classes as natural.
In Gramscian view, this hegemony of the ruling classes is a product of an active encouragement of subordinate classes to adopt a worldview that aligns with the interests of the ruling class. This worldview is continuously reinforced through educational institutions, the media, religious institutions and the wider sphere of civil society, all of which work towards a normalization of the values and priorities of the elite, making them appear inevitable and universal. The result of this normalized hegemony is that the elite secure their own existence while removing the thought of possible alternatives by simply defing the boundaries of what we perceive as possible, plausible, logical, reasonable or morally justfiable. This leads individuals to justify the existing social hierarchy as “the way things are, can or must be”, thereby stabilizing the power of the dominant classes without the need for overt repression. No one will resist the existing order if resistance has become unimaginable?
Here, we will attempt to examine how Gramsci’s concept of hegemony helps explain how powerful AI actors today use culture, technology, knowledge, and algorithms to shape what people accept as “normal”. It should still be noted that applying his ideas too directly to AI may carry risks. Unlike early 20th century class structures and struggles, that were somewhat localized, both the elite and technology are globally distributed and contested both locally and globally. It is therefore somewhat problematic to treat Big Tech as a unified "ruling class” with coherent ideological strategy. One might be able to argue that while Gramsci illuminates how perception and consent are shaped by the elite, they may not fully capture the complexity of decentralized major AI actors, and open source movements. Another argument against a literal adaptation of his ideas is that it may oversimplify a domain where scientific discovery, regulatory or technological forces interact in unanticipated ways.
To be fair, one could also argue that machine learning techniques, computation and alignment challenges cannot be reduced to ideology and that the view of AI as a hegemonic manipulation of the masses is far-fetched. Here, we argue that this line of reasoning is too simplistic and risks derailing the debate. Because it is not the AI models per se but the AI actors,it is not the researchers but the investors, and finally, it is not the technological advancement or new algorithms but the corporations that define its objectives of what the models are used for.
As such, the rapid expansion of AI and its deployment across the globe may have been organic in its early stages but this is no longer the case where major AI actors are the only ones that are capable of deploying really impactful AI systems. And as if they were a unified ruling class, they use their immense capital and “need for competitive edge” as a justification to push for deregulations, and postponement of various laws and penalties (EU AI Act delays) all the while they consume public and private data to build models whose applicability remain unforeseen and undeclared.
To make the matter more complicated, artificial intelligence itself is offered as the ultimate, inevitable and indispensable tool that will help resolve longstanding issues faced by humans. While it is reasonable to assume that AI systems will offer significant capital gains for the few, not to mention mass unemployment and other societal issues that the governments do not seem to have any plans for, AI is marketed as hope. The ruling classes shape the discourse on one hand as AI being capable of unimagined technical and scientific breakthroughs, and on the other hand as a means of creating a superintelligence that like the oracle of the myths, will offer some kind of meaning to the lives of one-dimensional beings under capitalism.
In this sense, AI development is an arena where the ruling classes and technocratic elites exercise hegemony through corporations and states by defining and negotiating the terms of public understanding in order to make their preferred technological order appear as natural and unavoidable. Indeed, from a Gramscian perspective, contemporary AI development can be understood as a new frontier for achieving cultural hegemony. Big Tech shapes not only markets but also the very interpretation of technology by average citizens. For instance, while the delays in the EU AI Act was motivated by the absolute need for competitiveness, innovation and progress, the influence and role of AI, despite the companies breaking the EU’s own regulations for a longer time than planned, is depicted as inevitable if not a necessity.
AI under capitalism is not a scientific algorithm in service of people. It not only serves the monetary interests of the few but also reflects power structures of the past and present with no incentives to create wealth for the masses. There are currently multiple AI systems embedded within corporate or state institutions that entrench existing inequalities by optimizing for profit, surveillance, or social stability in ways that seem “rational” but often are deeply political. A parallel to this in Gramsci’s world, is his warning about the intellectuals who reproduce the dominant ideology; As we progress, the models themselves may become the ultimate organic intellectuals, shaping culture at scale while appearing neutral. Masses can be manipulated through subtle forms of algorithmic pedagogy (ranking, summarizing, filtering, recommending etc). It is a matter of taste to consider the AI algorithms creating a dystopian mind control but they certainly can and do contribute to a gradual shift in what is considered thinkable or discussable.
Note that while both Gramsci and the Frankfurt school understood culture as a battleground for power, they emphasize different aspects of the problem. While the former focused on counter-hegemoney, the latter stressed the totalizing and homogenizing force of mass media. In fact, certain impactful AI models (LLMS) can be viewed as an extension of a culture industry, producing material that appears pluralistic while being structurally guided by opaque optimization and ill-understood objective functions. It seems quite clear that AI, unlike traditional media, is capable of reconfiguring the very conditions of resistance, deciding what becomes visible or thinkable. But AI’s technical or theoretical shortcomings (we do not know the actual objective functions), or the predatory abuse of the current techniques and computational possibilities aside, the crux of the matter is not the AI models themselves but the political economy within which the models are deployed to directly or indirectly automate the hegemony of the elite in historically unprecedented scales.
CounterCritical