The Machine Question is an independent research project exploring artificial intelligence through philosophy, technology and human cognition. Essays on intelligence, creativity and the evolving relationship between humans and machines.
Power
Who controls artificial intelligence?
Artificial intelligence is often discussed as if it were an autonomous technological force. In practice, every system depends on infrastructures, capital, data, energy, institutions and decisions that determine who can build it, who can access it and whose interests it ultimately serves.
The Question
Technology does not distribute power evenly.
Artificial intelligence may appear as software, but its capabilities depend on concentrated resources: computing infrastructure, capital, data, technical expertise, distribution platforms and the authority to determine how systems are deployed.
?
Where does control reside?
Discussions of artificial intelligence often focus on what a model can do. Yet capability alone explains little about who can build that model, who can afford to operate it, which data shape its behaviour or which institutions decide where it may be used.
The infrastructure of AI is unevenly distributed. Advanced systems depend on large concentrations of computation, energy, capital, specialised labour and access to global distribution channels. These dependencies create forms of power that may exist before any individual algorithm makes a consequential decision.
The central question is therefore not only whether AI changes power relations after deployment. It is how power is already embedded in the conditions under which artificial intelligence becomes possible.
The Conceptual Turn
The most consequential power of an AI system may not reside in any single output. It may reside in the capacity to define the infrastructure, standards and dependencies through which other actors must operate.
Three Distinctions
Capability, control and dependency are different forms of power.
Capability
A system possesses the technical capacity to perform tasks, influence decisions or transform processes. Capability creates potential power but does not determine who controls its use.
Control
Control concerns who sets access conditions, chooses deployment contexts, determines system constraints and retains the authority to change or withdraw the technology.
Dependency
Power can emerge when organisations, institutions or individuals become unable to function competitively without infrastructures controlled by another actor.
The Inquiry
The Machine Question therefore asks not simply who uses artificial intelligence, but who acquires the power to define the environments in which everyone else must use it.
What We Examine
Power begins before the output.
The field examines the structures that precede individual AI decisions: who owns the infrastructure, who determines access, which standards become unavoidable and how dependencies accumulate around systems that others do not control.
Who controls the conditions of computation?
Advanced artificial intelligence depends on chips, data centres, energy, networks and cloud infrastructure. Control over these resources can determine who is able to develop, scale and deploy powerful systems before questions of software capability even begin.
Who owns the systems others depend upon?
Ownership determines who can change a model, withdraw access, set commercial terms or capture the value created around it. The field examines how property rights and corporate structure shape technological power.
Who is allowed to use capability?
Power also resides in the ability to decide who may access a system, on what terms and with which limitations. Unequal access can create asymmetries between firms, institutions, states and individuals even when the underlying technology is theoretically general-purpose.
When does one architecture become the default?
Standards, interfaces and dominant platforms can shape entire ecosystems. Once organisations design processes around a particular infrastructure, technical choices can harden into institutional dependencies that are difficult to reverse.
Who defines the permissible?
Governments, firms, professional bodies and technical organisations all participate in setting the rules within which AI operates. Governance determines not merely what is forbidden, but which risks, values and interests become formalised in system design and deployment.
When does usefulness become dependence?
Technologies acquire structural power when abandoning them becomes costly or practically impossible. The field examines how convenience, integration and competitive pressure can transform optional tools into infrastructures that institutions can no longer easily refuse.
The Central Distinction
Influence is not the same as power. An AI system may influence a decision, but structural power arises when actors control the resources, standards and dependencies that determine which decisions can be made in the first place.
The architecture of AI is also an architecture of dependency.
Artificial intelligence is not deployed into an empty social space. Organisations reorganise workflows, skills, investments and expectations around technological systems. Each layer of integration can increase the cost of exit and the leverage of those who control the underlying infrastructure.
The field therefore studies not only visible decisions and regulatory conflicts, but the quieter process through which technical systems become conditions of participation in economic, institutional and cultural life.
Key Questions
Questions that reveal the structure.
The politics of artificial intelligence cannot be understood only by examining individual decisions or harmful outputs. The deeper questions concern who controls the conditions of participation, which dependencies become unavoidable and which forms of authority emerge from technological scale.
Can technological infrastructure become a form of political power?
When essential economic and institutional processes depend on privately controlled computational infrastructure, technical ownership may acquire consequences traditionally associated with public authority. Where, then, does technological capacity end and political power begin?
What happens when access to intelligence becomes a competitive resource?
If advanced AI substantially increases organisational capability, unequal access may produce cumulative advantages. Those already able to acquire superior systems may become better positioned to acquire the resources required for the next generation of systems.
Who governs systems that operate across jurisdictions?
AI infrastructures cross national borders while legal authority remains territorially organised. The field asks how states, corporations and international institutions negotiate authority when technological systems operate at a scale that no single jurisdiction fully controls.
When does a technical standard become an institution?
Standards can begin as engineering choices and end as conditions of participation. Once markets, organisations and public systems organise themselves around a technical architecture, changing that architecture may become economically or institutionally prohibitive.
Can dependency exist without coercion?
Actors may freely adopt a useful technology and nevertheless become structurally dependent upon it. Power may therefore arise without explicit force: through convenience, integration, network effects and the increasing cost of choosing differently.
What happens when power becomes difficult to see?
The most durable forms of technological power may become ordinary infrastructure. Once systems are experienced as necessary, neutral or simply given, the choices embedded within them can disappear from view even while their effects continue to structure behaviour.
Open by Design
The field remains open because power rarely resides in a single actor or decision. It emerges through relations among capital, infrastructure, institutions, standards and dependency — and often becomes strongest precisely when those relations begin to look inevitable.
Essays
Arguments developed through the field.
Essays extend the Power field into sustained arguments about infrastructure, concentration, dependency and the institutional conditions through which artificial intelligence acquires economic and political significance.
Featured Essay
The Infrastructure of Intelligence: Power Before the Algorithm
What if the most important power of artificial intelligence is created before a model produces a single answer?
The essay examines the layers beneath visible AI systems: compute, capital, data, energy, distribution and standards. Its argument begins from a structural premise: the actor who controls the conditions under which intelligence can be produced and accessed may acquire power independently of any particular algorithmic decision.
Explore the essays
Developing Research
Future essays will examine how technological usefulness becomes dependency and how states, institutions and firms respond when access to advanced intelligence is controlled through infrastructures they do not own.
The politics of dependency
An investigation into how voluntary adoption, convenience and integration can transform technological tools into infrastructures that organisations become unable to abandon without significant cost.
Sovereignty in the age of artificial intelligence
A study of what technological sovereignty can mean when states depend on privately controlled models, cloud infrastructure, semiconductors and transnational platforms for increasingly essential capabilities.
The Wider Corpus
Power also intersects with essays classified under Knowledge, Creativity, Intelligence and Humanity.
Explore all essays
Concepts & Frameworks
The ideas and lenses behind the field.
Technological power is rarely visible in a single decision. Concepts identify the forms through which control becomes structural; frameworks determine whether that power should be examined through infrastructure, institutions, markets or relations among states.
Concepts
What must be defined.
These concepts describe the mechanisms through which technological capability becomes durable influence over other actors, institutions and systems.
01
Control
The capacity to determine access, conditions of use, technical constraints, deployment contexts and the continued availability of a system.
02
Dependency
A relation in which abandoning a technological system becomes costly enough that formal freedom of choice no longer implies practical autonomy.
03
Concentration
The accumulation of capital, infrastructure, data, expertise or distribution capacity among a limited number of actors able to shape the field.
04
Sovereignty
The practical capacity of an institution or state to make consequential technological choices without depending entirely on infrastructure controlled elsewhere.
Frameworks
Where power is located.
Each framework reveals a different layer of technological power and prevents visible algorithmic behaviour from being mistaken for the whole system.
01
Infrastructural
Examines control over compute, chips, energy, networks, cloud systems and the physical resources on which advanced artificial intelligence depends.
02
Institutional
Studies how governments, corporations, professional bodies and standards organisations define rules, legitimacy and permissible uses.
03
Economic
Examines capital concentration, network effects, competitive advantage, switching costs and the distribution of value created through AI systems.
04
Geopolitical
Places artificial intelligence within relations among states, supply chains, strategic resources, technological blocs and competing claims to sovereignty.
Research Architecture
These connections are selective rather than exhaustive. The wider Concepts and Frameworks sections place Power within the broader architecture through which The Machine Question examines intelligence, language, creativity, knowledge and the human condition.