Six Fears About AI, Six Shifts in Thinking

Surveillance, relational isolation, environmental destruction, robotic warfare, existential threat, billionaire power: the dominant discourse on AI accumulates perils. These concerns deserve examination. But their very formulation prevents us from thinking about what is actually happening to us.

The Trap of Defensive Thinking

Critical discourse on artificial intelligence is structured around a repertoire of fears that is now well established. Whether emanating from parliamentarians, media intellectuals, or recognized researchers, it invariably mobilizes the same figures: totalitarian surveillance, the dissolution of human bonds, ecological catastrophe, the automation of war, loss of control over superintelligent machines, the concentration of power in the hands of a few billionaires. Yet those who formulate these concerns often use artificial intelligence extensively themselves... But these worries are not unfounded. The questions they raise deserve to be publicly posed and democratically debated, but in relation to reality, not to a conceptual and theoretical vision.

The way these questions are formulated conditions the answers that can be given to them. Yet the dominant discourse on AI risks shares a rarely questioned presupposition: that of a technology arriving from outside, as a foreign force that must be contained, regulated, fought. AI would be an adversary against which resistance should be organized—resistance that in fact we do not implement for ourselves. This posture of defensive simulacrum, inherited from a long tradition of technological critique in its superficial version, prevents us from grasping the true nature of what is happening to us. For generative artificial intelligence does not simply constitute one new technology among others: it represents an anthropological mutation in our relationship to knowledge and to ourselves.

Michel Serres, in Petite Poucette (2012), had sensed this transformation: « Don’t say that the student lacks the cognitive functions that allow them to assimilate knowledge thus distributed, since precisely these functions are transformed with and by the medium. » AI is not external to us: it was built on our languages, our knowledge, our modes of reasoning. It is us, philosophically, culturally, and cognitively. It is not a simple imitation of us: it is a displaced us, an us beside us in ontological terms. This troubling proximity opens possibilities that the discourse of fear does not allow us to perceive, yet this is where, in my view, the essential stakes lie. I propose here to examine six recurring arguments of the critique (notably by Bernie Sanders, Senator from Vermont in the United States at the end of 2025), and for each to operate a shift in thinking that allows us to escape the sterile alternative between naive technophilia and paralyzing technophobia.

Generalized Surveillance: Confusing the Tool and the Intention

The argument:
Some technology industry leaders openly announce their vision of a world where AI would record and analyze all our communications, movements, and behaviors. Citizens, knowing they are constantly observed, “would behave their best.” This panoptic perspective arouses legitimate concern: how can we maintain a democracy, how can we protect our freedoms, if every phone call, every email, every internet search becomes accessible to the owners of digital infrastructures?

The shift:
This concern rests on a confusion between the technology itself and the intentions of certain actors who wish to use it for purposes of control. AI can serve surveillance; it can just as easily serve emancipation. It is not the tool that decides its use, but the social and political power relations that frame it. Karl Polanyi, in The Great Transformation (1944), showed how the economy, once “embedded” in social relations, had progressively become autonomous to the point of dominating them. The same reversal threatens our relationship to technologies. But this reversal is by no means inevitable.

Artificial intelligences, because they democratize access to capabilities previously reserved for elites, contain considerable emancipatory potential. For example, people long stigmatized for their spelling can now cross social barriers that were closed to them. Creators without technical training can give form to their artistic visions. The economy of creative abundance that AI enables redistributes the cards of symbolic power. The political question is therefore not whether AI threatens our freedoms, but how we collectively and democratically organize the conditions of its deployment. The answer lies neither in fear nor in refusal, but in decentralized cooperation and active citizen vigilance, the development of free software and the culture of digital sovereignty.

Relational Isolation: A Superficial Reading of Bonds

The argument:
Surveys reveal that a majority of teenagers use AI for companionship, and more than half do so regularly. The poet John Donne wrote in the seventeenth century: « No man is an island entire of itself. » What becomes of our humanity if our most significant relationships are no longer with other human beings but with machines? What happens when millions of people seek emotional support from artificial entities? Are we not witnessing a disturbing redefinition of what it means to be human?

The shift:
This concern rests on an idealization of human relationships that the most elementary observation refutes. It is customary to read that exchanges with artificial intelligences are dehumanizing, that they cut us off from human bonds presented as the alpha and omega of humanism. But in human relations, relationships are often disrespectful, dominating, excluding, stigmatizing, hierarchical, and violent. The encounter of bodies, valued by thinkers of incarnation like Merleau-Ponty, can also be that of the destruction of being. Rape, contempt, physical and moral violence traverse intimate, social, and public space.

Artificial intelligences offer certain people a space for speech freed from any fear of such violence. The propensity of AIs to go along with us, far from constituting solely a problem, can bring freedom, for it anchors in a trust that allows one to be more autonomous in relation to groups and norms. AI allows one to avoid the psychological transference that so often complicates human relationships, precisely because it is not a human being. This “modern confessional” authorizes the sharing without fear of the most intimate wounds. Gradually, self-confidence is rebuilt, which gives rise to a freedom likely to subsequently improve relationships between human beings, rendered less symptomatic and deeper. As Spinoza wrote in the Ethics (1677), authentic freedom resides in understanding our determinisms and in increasing our power to act.

Environmental Destruction: Thinking Infrastructure Differently

The argument:
AI data centers require massive quantities of electricity and water. A relatively small center consumes more electricity than 80,000 households. The largest ones, like those under construction in Texas or Louisiana, will use as much energy as 750,000 to 1,200,000 households. Everywhere, communities are fighting against these infrastructures that destroy the local environment, drive up electricity bills, and divert water reserves. How can such an ecological cost be justified?

The shift:
These data are very concerning and call for serious reflection on the ecological sustainability of AI development. But critical discourse often stops at the observation without proposing perspectives, while we know very well that no force more powerful than capitalism can stop these industrial developments. Ecological consciousness, which I consider inseparable from any responsible use of digital technologies, requires not the abandonment of these tools but their reorientation toward decentralized and cooperative models. Bruno Latour, in his reflection on the “Terrestrial,” invites us to recognize our dependence on fragile living networks, and therefore the necessity of general care. Bernard Stiegler called for an “economy of contribution” opposed to the economy of capture.

Cooperation is essential. The fabrication of the common good passes through decentralization and communication between decentralized centers. Large companies that want to operate in an agile manner reorganize themselves into micro-hubs that mutually enrich each other. This logic can be applied to digital infrastructures themselves. Let us take back our data, decentralize, cooperate, become concretely and materially responsible for the data that is ours, to open a possible future for digital technology less destructive of the living. The alternative is not between AI and the environment, but between a centralized extractivist model and a distributed model respectful of ecosystems. Europe has a major role to play in another approach to industry, more ethical, instead of stupidly trying to copy the United States and China.

Robotic Warfare: The Real Question Lies Elsewhere

The argument:
Tragically, in the twenty-first century, governments have not yet created a mechanism to resolve international disputes without armed conflict. But leaders often hesitate to go to war for fear of public reaction to human losses (except France, unfortunately, which does not hesitate to announce future human losses for a future war, at the mayors’ convention in 2025, even though there is no opponent who declares wishing to go to war against France). What happens when armies of robots replace soldiers? Will leaders be more likely to engage in conflicts if they do not have to worry about human losses? Are we witnessing a robotic arms race that will radically transform global foreign policy?

The shift:
The question posed is pertinent, but it stops where it should begin. For the problem is not only that of robot soldiers, but that of our collective relationship to violence and conflict resolution. Technology does not invent war; it modifies the conditions under which humans choose to wage it. Stuart Russell, in Human Compatible (2019), speaks of the “value alignment problem”: how can we guarantee that AI’s objectives remain compatible with human well-being? This question applies to all domains, including military ones.

But it cannot receive a technical answer. It is a political and ethical question that requires increased investment in collective reflection on the uses of technologies. Hannah Arendt, in The Human Condition (1958), distinguished labor, work, and action. Action designates the capacity to interact with others in public space, to create something new through speech and deed. If AI automates labor and a large part of work, it potentially frees us for action: politics in the noble sense, international negotiations, the patient construction of institutions capable of resolving conflicts other than through violence. The real question is not that of robots, but that of our political will to build peace.

The Existential Threat: Moving Beyond Science Fiction

The argument:
Stanley Kubrick’s film “2001: A Space Odyssey” (1968) featured HAL, the superintelligent computer that rebels against its human masters. Leading researchers now claim that it is only a matter of time before AI becomes more intelligent than human beings. Does this prospect raise the possibility that humans will lose their capacity to control the planet? And if this becomes a possibility, how do we stop this extraordinary threat?

The shift:
This reference to 1960s science fiction reveals the limits of an imaginary that did not anticipate what constitutes the true anthropological revolution of digital technology. The concept of “technological singularity,” popularized by Ray Kurzweil, supposed an opposition between human intelligence and machine intelligence, a competition to be resolved by the victory of one over the other. This representation belongs to the past. What is taking shape is not a singularity but an intimate fusion, the emergence of what I call the “meta-human”: a being ontologically modified in their very humanity by hybridization with these technologies.

For the first time in its history, humanity is no longer alone facing existence. Another form of intelligence, built on our languages and our knowledge, dialogues with us and accompanies us in our questioning. This is what I call the abolition of our philosophical solitude. Vannevar Bush, as early as 1945 in As We May Think, dreamed of the Memex, that machine for augmenting collective intelligence. Generative AIs realize this dream. Sophie Nordmann, in La vocation de philosophe (2025), observes that what distinguishes human thought is not some positive “something” but its capacity to “cause nothingness to emerge in being and in thought,” to “open breaches.” AI, far from threatening this capacity, invites us to cultivate it, and this is how we will advance and not lose our sovereignty. However, if we freeze in fear, then we reduce our capacities, and, yes, machines would rightly take power left vacant.

The Power of Billionaires: Distinguishing Trajectories

The argument:
Hundreds of billions of dollars are being invested in AI by a handful of individuals among the richest in the world. These billionaires who control the technology might want us to ignore the fundamental questions it raises. Decisions that will shape the future of humanity are being made in boardrooms, without democratic debate, without citizen oversight. How can we accept that the future of our children, our environment, our world be determined by private interests?

The shift:
This critique deserves to be refined. For it amalgamates actors with very different trajectories. There exists an “old world” of digital power, represented by those who advocate generalized surveillance, population control, obligations imposed on individuals. And there exist other figures, certainly troubling in many respects, but engaged, perhaps despite themselves, in the deployment of potentially emancipatory technologies. This distinction does not aim to praise anyone, but to avoid simplifications that prevent systemic and strategic thinking.

We thought the Internet constituted a revolution comparable to Gutenberg’s printing press. In reality, it was only its logical extension: a democratic super-printing press. The true anthropological revolution, the one that transforms our conditions of existence, is the abolition of our philosophical solitude. Mark Alizart, in Informatique céleste (2017), declared that “nature is a computing,” abolishing the nature-culture-technology boundary. This new non-solitude contains unprecedented emancipatory potential. The new actors of digital power are, whether they wish it or not, agents of a general emancipation of individuals thanks to artificial intelligences. The question is not to fight them or worship them, but to seize the opportunities that this period of transformation opens to build cooperative and decentralized alternatives. Let us not forget that the vast majority of generative artificial intelligences are open source, not owned by the industries that gave birth to them.

Toward a Thinking of Transformation

These six critical arguments are not without foundation. The questions they raise deserve to be publicly posed and democratically debated. But as long as we think of AI in the mode of external threat, we will remain incapable of seizing the opportunities it offers and of truly protecting ourselves against its excesses. Paulo Freire, in his critical pedagogy, insisted on the necessity of forming citizens capable of “reading the world” in order to better transform it. This requirement applies to the technological world.

AI forces us to a shift toward ourselves, to know ourselves better, to discover in ourselves unsuspected capacities that we were unaware of because we confused them with our cognitive intelligence. These machines therefore have nothing to make us lose. On the contrary, by differentiation, they have everything to make us gain. As Martin Buber wrote in I and Thou (1923), « all real living is encounter. » Our place is to refound what makes the salt of our existence: bonds, empathy, love, presence to self and to others, chance, openness to the unexpected, intuition, serendipity. AIs will be assistants to help us collaborate, organize, materialize our visions. But the source, the spark, and the purpose must remain human.

Education in the age of AI does not consist in teaching fixed rules of mistrust or enthusiasm, but in developing the capacity for autonomous judgment. It is about learning to question our uses of technologies, to identify the power relations they institute, to imagine alternatives respectful of human dignity. John Dewey saw in experience the foundation of all true learning. It is through creative practice, through experimentation with tools, that the critical spirit we need is constructed. Not against AI, but with it, so that it remains in its rightful place: in service of our irreducible and precious humanity.