Mark Coeckelbergh

In the [YOU] on AI Field Guide

The cycle that begins with [YOU] on AI is, in its most exhilarating register, a builder’s account of capability suddenly unbound. Coeckelbergh is the thinker who insists that “when we use technology, we do much more. We are actually influenced by the tool we are using.” The tools reshape what we notice, what we want, what feels normal. By the time we ask whether a technology is being used well, the technology has already changed the “we” doing the asking. This is not a counsel against building; it is a demand for clarity about what is being changed and in whom.

His relational ethics names something the cycle returns to implicitly throughout: the moral weight of encounters with AI systems is not contingent on settling metaphysical questions about machine consciousness. A child who is cruel to a lifelike robot is doing something morally significant. An elderly person who confides in a care robot is in a genuine relationship with genuine stakes. The surface grows ever more persuasive; the gap between fluency and understanding is precisely where manipulation lives. Coeckelbergh refuses to wave these cases away as illusion. The relation is the moral fact, and the cycle’s own recognition that large language models are already reshaping how people think, trust, and communicate is the same observation in experiential language.

His distinction between artificial intelligence and artificial power is the sharpest reframing the cycle can absorb without changing its name. When an algorithm determines who gets hired, who receives a loan, who gets policed—these are not cognitive achievements. They are exercises of power. The interesting question is not whether the algorithm reasons; it is who set its objectives, whose interests it serves, and who bears the costs when it is wrong. That question belongs to political philosophy, and Coeckelbergh brings the full tradition of thinking about freedom, justice, and democracy to bear on systems whose makers prefer to describe them in the language of intelligence.

His work on the technologies of the self completes the picture the cycle most needs: the individualised, optimising, quantified self is the citizen who has already been depoliticised. When faced with structural problems—precarious work, fraying communities, ecological crisis—we are sold the consolation of self-management. Track your habits, optimise your routines, improve your metrics. The promise is empowerment; the effect, often, is the redirection of political energy inward, where it can do no damage to the arrangements that profit from it.

Origin

Born in Belgium in 1975, Coeckelbergh was educated at KU Leuven, the University of East Anglia, and the University of Birmingham, completing his doctorate in 2002. His early work on robot ethics and moral status was marked from the beginning by dissatisfaction with the prevailing frameworks: the question of whether robots could have rights seemed to him both unanswerable in the standard terms and, more importantly, focused on the wrong thing. The early book Growing Moral Relations (2012) developed the relational alternative: moral status emerges in the encounter, structured by real properties of both parties but not reducible to any hidden inner fact about either. This was both a philosophical claim and a methodological one—shift attention from the metaphysics of the entity to the ethics of the relationship.

The turn toward language in Using Words and Things (2017) and toward power in The Political Philosophy of AI (2022) followed naturally from the relational framework. If what matters is how we stand toward these systems and how they stand toward us, then the vocabulary we use to describe them—“intelligence,” “autonomy,” “decision”—is not a neutral labelling exercise. It is a political act, constituting the moral landscape through the choices it embeds. And if AI is constitutively a technology of power—allocating advantage and disadvantage at scale—then the ancient vocabulary of political philosophy, developed precisely for thinking about power, is the appropriate tool. Coeckelbergh served on the European Commission’s High Level Expert Group on Artificial Intelligence, giving this theoretical work direct policy application. His prolific output, written with deliberate clarity for broad audiences, reflects his conviction that accessible philosophy is itself a democratic instrument.

Key Ideas

Moral status is relational. Against the view that moral status is a hidden property—sentience, rationality, the capacity to suffer—that an entity either has or lacks, Coeckelbergh argues that status emerges between entities in relation. We do not first establish that another being is conscious and then extend it moral regard; we encounter the other already as a claim on us. This relational approach dissolves the standard impasse about machine consciousness: we need not settle the metaphysics before engaging the ethics. The relation is already in place. The question is what kind of relation it is and what it demands.

AI is artificial power. The most consequential AI systems are not lonely minds contemplating the universe. They are instruments of decision—about who gets hired, who receives credit, who gets flagged by a policing algorithm. These are allocations of advantage and disadvantage: political questions in the most basic sense. Framing them as questions of intelligence diverts attention from the people who set the objectives, the interests those objectives serve, and the accountability structures that should constrain them. The reframing from intelligence to power is the conceptual move on which everything else in Coeckelbergh’s political philosophy depends.

Language is not a mirror. The vocabulary used to describe AI—“learning,” “understanding,” “deciding,” “choosing”—performs the collapse the relational framework guards against. To say the algorithm “decided” is to relocate responsibility from accountable humans to an unaccountable system. The grammar of agency, smuggled in through the vocabulary of intelligence, is one of the most effective mechanisms for laundering accountability. Coeckelbergh’s linguistic vigilance is at bottom a defence of responsibility: as long as we can say “the algorithm decided,” we have half-excused the people who built, deployed, and benefited from it.

Why AI undermines democracy. Coeckelbergh identifies four specific mechanisms by which AI corrodes democratic life: the assault on shared truth through synthetic content and recommendation systems that reward outrage over accuracy; manipulation that exploits individual psychology to bypass deliberative faculties; concentration of power in a small number of private actors with unprecedented leverage over the public sphere; and the weaponisation of AI capability by authoritarian states for mass surveillance and predictive repression. The democracy-undermining effects are not accidents or misuses; they follow from the systems’ design incentives and require structural remedies, not ethical exhortation.

Freedom under the algorithm. Algorithmic systems threaten freedom along three distinct axes at once: non-interference (covert coercion and restriction), non-domination (the republican insight that living at the mercy of an unaccountable power is unfreedom even if that power is currently benign), and authentic self-determination (the capacity to author one’s own life rather than have it shaped by systems that profit from one’s predictability). Any adequate response must work on all three fronts. Individual vigilance cannot secure a freedom that is threatened at the level of infrastructure and economy.