Artificial intelligence enters the history of executive power as a factor of acceleration and structural reorganization, not as a simple administrative innovation. In contemporary institutional systems, the capacity to collect, process and interpret information in real time alters the relationship between decision, control and legitimacy. The adoption of predictive systems and advanced analytics platforms does not only affect the efficiency of public action, but redesigns the balance among the powers of the state, systematically strengthening the position of the executive.
The first observable effect is the increase in the operational capacity of central government. Continuous monitoring systems, macroeconomic simulation models and risk assessment tools enable executive leadership to intervene with a speed incompatible with the traditional timing of parliamentary deliberation. During the pandemic, the use of epidemiological models and tracking platforms allowed many governments to adopt restrictive measures and allocate health resources in extremely short timeframes. According to data from the World Health Organization, more than 80 countries used advanced modeling systems to support emergency executive decisions, often through task forces directly connected to the highest levels of government. This type of decision making infrastructure strengthens the executive not through formal legal means, but through a permanent informational advantage.
The centralization of operational knowledge is the corollary of this transformation. Data flows and the models that interpret them are concentrated in a limited number of specialized units, often located within the executive or under its direct control. Those who possess dashboards, simulations and real time indicators exercise a form of power that precedes formal decision making. Parliaments, independent authorities and intermediate administrative levels often receive already aggregated information, difficult to contest on substantive grounds. According to a 2023 OECD report, in more than 60 percent of member countries governmental data analytics systems are managed by structures directly dependent on the executive, with limited access for other institutional powers.
This informational asymmetry produces a progressive marginalization of intermediate bodies. This is not an explicit displacement, but a functional erosion. If decisions are presented as technically optimal responses to complex scenarios, the role of political and social mediation loses effectiveness. Parliaments debate measures already strongly preconfigured by models, while oversight authorities struggle to exercise effective supervision over opaque and highly technical processes. The result is a silent transformation of democratic checks and balances, which risk being reduced to mechanisms of ex post legitimation.
From an institutional perspective, this dynamic places the principle of separation of powers under pressure. Not through authoritarian overreach, but through a practical redefinition of competences. Executive power accumulates not only decision making capacity, but anticipatory capacity. In a context in which prediction becomes a strategic resource, those who control the models also control the agenda of possible choices. This is a form of power that operates upstream of the political process, delimiting what appears realistic, sustainable or urgent.
The economic implications are significant. Decision making concentration favors policies oriented toward short term stability, risk reduction and optimized resource management. According to the International Monetary Fund, the use of predictive systems in fiscal policy has contributed in some countries to improving public spending efficiency by up to 15 percent. However, the same analysis highlights increased decision making rigidity and reduced capacity to adapt to non modeled shocks. Algorithmic efficiency tends to privilege what is measurable and predictable, penalizing redistributive policies or long term investments with uncertain outcomes.
Hannah Arendt warned against the reduction of politics to technical administration. When power is exercised primarily through process management, public confrontation is replaced by optimization. In the era of artificial intelligence, this intuition takes on a new form. The risk is not declared authoritarianism, but a governance that operates by default, guided by indicators and forecasts, in which political responsibility dissolves into the complexity of systems.
The parallel with the corporate world reinforces this interpretation. McKinsey studies show that companies that excessively centralized strategic decisions through algorithmic systems achieved short term efficiency gains, but displayed greater vulnerability to sudden shocks, precisely due to the reduction of bottom up feedback and decision making diversity. Similarly, a state that strengthens the executive through AI may appear more reactive, but risks becoming less resilient at the institutional level.
The central issue does not concern whether to use artificial intelligence in government action, but how it is incorporated into the constitutional and administrative architecture. A stronger executive is not in itself a problem. It becomes one when such reinforcement is not accompanied by new instruments of control, transparency and accountability adequate to the complexity of the systems employed. Without an adaptation of democratic checks and balances, AI tends to amplify already existing centralizing tendencies.
In a context in which command can be exercised through invisible infrastructures and predictive models, the challenge for institutions is not only to decide more quickly, but to preserve the capacity to answer for decisions. The transformation of executive power in the era of artificial intelligence thus opens a field of tension destined to mark for a long time the relationship between technology, economy and democracy.
Global AI Observatory