Contemporary technocratic authoritarianism: expertise, performance and government by data

Series: Strategic Foundations

Authors: La Ligne Fine – Institut

Published Date: October 15, 2026

Executive Summary

Authoritarianism is generally analysed from the point of view of the concentration of power, the limitation of political competition, coercion or the control of information. These dimensions remain essential, but they are not sufficient to account for certain contemporary transformations in the exercise of power. At the same time, several political systems are giving increasing importance to administrative expertise, the selection of elites, quantified objectives, performance evaluation, digital infrastructures and, now, artificial intelligence. Authority is not only capable of imposing a decision: it also seeks to demonstrate that it knows how to administer faster, plan longer and mobilize state resources more effectively.

This development comes in an international context of documented declines in several democratic indicators. According to the V-Dem Institute’s typology, at the end of 2025, 92 states fell into one of the autocracy categories in its ranking, compared to 87 democracies, and about 74% of the world’s population lived in regimes that this methodology classifies as autocratic. These figures depend on the definitions and thresholds used by V-Dem and therefore do not constitute an undisputed classification of each regime, but they do identify a general trend: the question of the institutional capacity of non-democratic systems now concerns a significant part of the world’s population.

However, it would be wrong to infer from this evolution the emergence of a homogeneous political model that could simply be described as « authoritarian technocracy ». Technocracy is not in itself a political regime. Democracies also entrust considerable responsibilities to specialised administrations, such as central banks, independent agencies, scientific experts or regulatory authorities. Similarly, the digitalisation of administration is not inherently authoritarian. The OECD documents the use of artificial intelligence by public administrations to automate or personalise services, improve decision-making, detect fraud and assist public officials, while highlighting the risks of bias, opacity and dilution of responsibility.

Technocratic authoritarianism must therefore be defined more precisely as a configuration in which the limitation of political contestability coexists with a strong valuation of the state’s expertise, administrative competence, measurable performance, and informational capacity. These elements do not replace coercion, ideology, or the personalization of power. They add to them and can provide the regime with additional resources for legitimation and government.

Three transformations must be distinguished. The first concerns legitimacy: economic performance, the quality of infrastructure, security, the speed of services or the ability to manage a crisis can be mobilized to justify the quality of a political system. The second concerns administrative capacity: the integration of databases, digital identity, government platforms and artificial intelligence potentially increase the amount of information accessible to the state and reduce the cost of certain interventions. The third concerns control: the same infrastructures can facilitate the identification, surveillance, censorship or anticipation of behaviors considered politically sensitive.

These three dimensions do not, however, automatically converge towards an omniscient state. On the contrary, they produce a fundamental contradiction. A technocratic regime seeks to govern with more abundant and accurate information, but a political system in which bad news is costly to convey, indicators determine careers, and contestation is limited can encourage administrative agents to filter or modify this information. The literature on China has documented the incentives to manipulate certain local statistics and the efforts of the central government to circumvent the distortions produced by its own bureaucratic hierarchy.

Data does not therefore eliminate the informational problem of authoritarian power. It can modify its scale, reduce certain manifestations and create new ones. The more a system measures performance, the more it is in the interest of actors to satisfy the metrics by which they are evaluated. The more information the state collects, the more decisive its integration, quality and interpretation become. The more automated systems participate in decision-making, the more important the ability to identify and correct their errors becomes.

The strategic challenge of contemporary technocratic authoritarianisms thus lies less in the hypothetical emergence of an entirely new « algorithmic dictatorship » than in the transformation of the relationships between knowledge, execution and control. Their potential advantage lies in the concentration of resources, the continuity of certain policies and the speed of implementation. Their vulnerability lies in the quality of the information going up to the centre, in the ability to correct a policy without institutionalised contestation and in the dependence of part of their legitimation on the results they claim.

The central question of this Paper is therefore the following: how do expertise, performance and data transform the mechanisms of government and legitimization of contemporary authoritarianisms, and what vulnerabilities does this technocratization produce in return?

Problem

Technocracy has a less obvious definition than it seems. In its broadest sense, it refers to the importance given to the holders of specialized knowledge in the formulation or execution of public decision-making. However, this definition encompasses extremely different institutional configurations. McDonnell and Valbruzzi have shown that the very notion of technocratic government remains difficult to stabilize in the literature and have proposed to distinguish technocrats from partisan politicians according to their trajectory, their relationship to parties and their professional legitimacy. Their study focuses on European democracies and reminds us precisely that technocracy and authoritarianism are not synonymous.

Expertise is in fact a normal component of the modern state. Fiscal, energy, military, health or monetary systems are complex enough to require specialized bureaucracies. A democracy can therefore be highly technicalized while maintaining political competition, parliamentary control, judicial remedies and the possibility of alternation. Conversely, an authoritarian regime can be weakly technocratic when it is based mainly on personal loyalty, patronage or coercion and administrative expertise occupies a secondary place.

The expression « technocratic authoritarianism » cannot therefore refer to a simple government of experts. It describes here a particular articulation between four mechanisms: the concentration of political authority; the valorization of expertise in recruitment and decision-making; the use of performance as a justification for public action; and the development of infrastructures for measuring, processing and exploiting information on a large scale.

This definition also makes it possible to distinguish between technocracy and meritocracy. Meritocracy is mainly concerned with the way in which individuals are selected and promoted according to criteria presented as objectives: exams, diplomas, professional results or administrative evaluations. Technocracy is more concerned with the political status accorded to specialized competence. The two can be reinforced: a meritocratic recruitment system can feed into a technocratic bureaucracy, which can then present its competence as a justification of its authority. But they remain analytically different.

A final distinction concerns digital technology. The digital state is not necessarily an authoritarian state, and authoritarianism obviously did not wait for digital technologies to monitor its population. Rather, contemporary transformation is about increasing the amount of information that can be collected, linked and processed. Administrative bases, digital identities, telecommunications, biometric systems, cameras, platforms and artificial intelligence can increase the « readability » of society for the state. The political question then is about the institutions that determine what can be collected, for what purpose, under what control and with what possibilities of recourse.

1. Technocracy is not a political regime

The confusion between technocracy and authoritarianism stems in part from their common relationship to a critique of political deliberation. Technocratic reasoning is based on the idea that certain decisions can be improved by expertise, measurement and analysis. It becomes politically problematic when an issue involving value arbitrations is presented as having a purely technical solution, making the challenge not only undesirable but irrational. The transition from expertise to political technocracy therefore occurs when the claimed competence tends to reduce the legitimacy of other forms of participation in decision-making.

This tension also exists in democracies. Independent central banks are the clearest example of the delegation of important decisions to institutions that are partly protected from the electoral cycle. Health agencies, competition authorities, energy regulators or constitutional courts also have varying degrees of autonomy. However, these institutions remain in an order where their mandate can be challenged legally or politically and their autonomy is defined by rules that are not entirely theirs.

In an authoritarian system, expertise can fulfil a different function when it becomes one of the arguments justifying the low contestability of power. The idea is no longer just that certain policies require specialists, but that the continuity, competence or supposed rationality of the ruling apparatus would produce better decisions than political competition. Technocracy then becomes a component of the representation that the regime gives of itself.

However, one should avoid retrospectively reconstructing all developmentalist states from this category. The economic bureaucracies of post-war Japan, South Korea, Singapore, or reforming China developed in different political systems and at different times. Their common thread relevant to this analysis lies less in their regime than in the emphasis placed on administrative capacity, planning, selection of cadres, and the use of measurable objectives.

Singapore is a particularly useful frontier case in this respect. A large body of literature analyses the place of meritocracy, administrative elitism and technocracy in the construction of the Singaporean model. Recent works describe the persistence of a « culture of meritocracy » associated with the political dominance of the People’s Action Party and a highly interventionist developmental state, while other studies insist on the Singaporean combination of elite selection, strong administrative capacity and the use of digital technologies. This literature contains divergent normative interpretations; the interest of the case is therefore not to artificially fit it into a homogeneous category, but to show that administrative competence, meritocracy and political concentration can be combined to varying degrees.

This first distinction is essential to understand the contemporary phenomenon. It is neither the expert, nor the indicator, nor the algorithm that makes a system authoritarian. The essential political variable remains contestability: who can question the decision, obtain the information that made it possible to make it, sanction its authors, propose an alternative policy and possibly replace the leaders?

1. Performance as a political resource

Political regimes are not maintained exclusively by coercion. They also seek to justify their right to govern. Tannenberg, Bernhard, Gerschewski, Lührmann, and von Soest distinguish four broad categories of legitimation strategies: ideological, personal, procedural, and performance-based. These mechanisms are not exclusive and can be combined by the same regime.

Technocratization particularly reinforces the last category. A power can present economic growth, infrastructure, security, education, public services or the effective management of a crisis as proof of its ability to govern. This logic transforms administrative performance into a political resource: efficiency is no longer sought only for its material effects, it also participates in the justification of the system that produces it.

This mechanism explains the importance given to indicators. An administration that presents itself as efficient must be able to materialize this performance. Growth rates, kilometres of railways, administrative delays, job creation, school results, the number of housing units, industrial investments or digital coverage all become measures that make it possible to tell the story of public action. Government by objectives thus has a dual function: to guide the bureaucracy and to make the result visible.

China has taken this logic of administrative management by objectives, evaluations and bureaucratic careers very far. But administrative meritocracy can fulfil an even broader function. Hanzhang Liu has studied the Chinese civil service national competition as a co-optation mechanism. His analysis concludes that the perception of a possibility of promotion through a relatively open and rules-based procedure can increase the sense of social mobility of the graduates concerned and reduce their demand for redistribution. The interest of this result is to show that meritocratic selection does not only act on the supposed quality of the bureaucracy: it can also integrate individuals into the system by making a prospect of progression within it credible.

Performance can also be constructed politically from a specific event. Bruno Schmidt-Feuerheerd has thus studied the way in which the management of the Covid-19 pandemic was mobilized in Saudi Arabia as a resource of legitimation. His work shows that this legitimization was not only the result of vertical government communication: the media and supporters of the government participated in the construction of a comparative narrative presenting the Saudi response as proof of a superior capacity of the state.

This observation introduces an important limitation. Performance and its representation are not identical. A regime can obtain results without succeeding in transforming them into legitimacy; it can also select the results it puts forward. A policy of legitimation through performance therefore always involves an informational dimension: producing indicators, choosing comparisons and building a relationship between the result obtained and the supposed qualities of the political system.

This dependence on performance can itself become a vulnerability. An ideological legitimacy can interpret material failure as a necessary sacrifice to a higher cause. A dynastic or religious legitimacy has other resources for continuity. A justification based primarily on competence exposes power more at the time when results deteriorate. The economic downturn, the failure of a major project, a poorly managed crisis, or the inability to maintain the level of services can then affect not only a particular policy, but the argument that the concentration of power produces a more efficient administration.

Technocratic authoritarianism is thus confronted with a particular requirement: it must continually produce, measure or stage the capacity it claims.

1. There is no single model of technocratic authoritarianism

The comparison between China, the Gulf monarchies and Singapore makes it possible to identify distinct configurations without assuming that they belong to a single institutional type.

China combines a ruling party, a hierarchical bureaucracy, mechanisms for selecting and promoting cadres, national plans, and extensive use of quantified objectives. The importance of expertise does not mean that political selection is based exclusively on technical criteria: political loyalty, the internal balances within the Chinese Communist Party, and the priorities defined by its leadership remain decisive. The term technocracy alone cannot therefore account for the functioning of the system. On the other hand, it makes it possible to identify the importance given to administrative capacity and knowledge in the official representation of the government.

China’s digital government strategy exemplifies this orientation. The State Council’s guidelines on the development of digital government set goals for improving so-called scientific decision-making, more precise social governance, and more efficient public services. The strategy explicitly linked data, administrative coordination, security, and modernization of government capacity.

This architecture has deepened with the creation of the National Data Administration and the increasing integration of public information resources. In 2026, Chinese institutional sources indicated that the national government data exchange system was linked to the platforms of 62 government departments and 31 provincial entities, and that the integrated government system had aggregated several million data catalogs. These figures come from the Chinese authorities and should therefore be understood as official administrative data, but they show the scale sought by the digital government project.

The Gulf monarchies have a different trajectory. Saudi Arabia and the United Arab Emirates have neither the Chinese party structure nor its bureaucratic heritage. Their technocratization is closely associated with economic transformation, diversification away from hydrocarbons and the construction of new specialized administrations. Planning here takes the form of national strategies with quantified objectives, investment programs and monitoring mechanisms.

The 2025 annual report of Saudi Vision 2030 explicitly presents artificial intelligence as an instrument to improve sector performance and describes the goal of a more data-driven government, using AI to support public decision-making. It combines this ambition with the localisation of technological capabilities and the construction of a national data and artificial intelligence ecosystem.

In the United Arab Emirates, the digital government strategy is officially based on eight dimensions, including digital by design, data-driven, open by default and the proactivity of services. Data is explicitly considered as a strategic asset intended to produce public value, plan policies and monitor their implementation. The interest of this case lies in the way in which administrative transformation, economic attractiveness and digital technologies are integrated into the same strategy for modernizing the state.

Singapore is on yet another trajectory. The Singaporean developmental state has historically been built around a professional bureaucracy, a strong valuation of competence and an explicit conception of meritocracy. Some of the literature analyses this organisation as a « political meritocracy » while other works place more emphasis on restrictions on political competition and the elitist dimensions of the system.

Digital transformation is now extending this administrative tradition. Singapore’s Ministry of Digital Development and Information presents the goal of digital government as simplifying transactions, improving accessibility, inter-ministerial coordination, and developing the digital skills of public officials.

These three trajectories show why the notion of technocratic authoritarianism must remain an analytical tool rather than a rigid institutional category. In China, technocracy is integrated into a centralized party system. In the Gulf, it accompanies an economic transformation led by monarchies with significant financial resources. In Singapore, it is part of the continuity of a developmental state, a political culture of meritocracy and particular electoral institutions. The mechanisms are comparable without being equivalent.

Their most relevant commonality lies elsewhere: they all place a high value on the state’s capacity to execute and seek, in different ways, to convert this capacity into a political resource.

1. Digital technology increases the government’s information capacity

The modern state has long depended on its ability to make society administratively legible. Censuses, cadastres, tax registers, civil status, economic statistics and administrative files make it possible to transform a complex population into categories that can be used for public action. Digital technology does not create this function; it modifies its speed, volume and possibilities of interconnection.

The first transformation concerns collection. Administrative interactions now leave digital traces that can be stored and analyzed. The second concerns integration: information previously dispersed between several ministries can be linked around common identifiers or exchange platforms. The third concerns processing: automated systems can search for anomalies, identify correlations or produce recommendations on a scale that is difficult for an exclusively human administration.

These capabilities have ordinary uses that are not subject to political control. The OECD identifies among the public applications of AI the automation of tasks, the personalization of services, decision support, fraud detection or the improvement of the work of agents. At the same time, it stresses that data quality, transparency, accountability and the maintenance of human supervision are decisive conditions for these uses.

Potential efficiency comes in particular from the reduction of coordination costs. A citizen who previously had to provide several administrations with information already held by the State may, in an integrated architecture, have to transmit it only once. A benefit can be awarded automatically when a change in situation is detected. An administration can identify inconsistencies between several databases without launching a manual investigation. A government can monitor certain economic or territorial indicators in almost real time.

China is explicitly seeking to develop this integration. The regulations on government data sharing adopted in 2025 aimed in particular to reduce administrative « silos » and build a national framework for exchange between administrations. The Chinese authorities present these systems as instruments for coordination, service delivery and economic monitoring.

The strategic issue is important because the capacity of the state does not depend only on the volume of resources it has at its disposal. It also depends on its ability to know where to allocate them. An administration with more information can theoretically target a policy more precisely, identify a failure more quickly and coordinate several institutions more effectively.

But this increase in readability also changes the relationship between the state and society. Information collected to award a service can potentially be used for another purpose. Interconnection reduces the silos that sometimes protected individuals simply because different administrations did not have the same data. The increase in administrative capacity is therefore simultaneously an increase in the potential capacity for control.

This is where the institutional difference becomes decisive. In a system with strong data protection rules, independent authorities, judicial remedies, algorithmic transparency and parliamentary oversight, the secondary use of data can be limited or challenged. When these mechanisms are weak, the same technical infrastructure can be used on a much wider scope.

Digital technology is therefore not authoritarian by nature. It increases the number of things a state can do. The political question is about what it is allowed to do and which institutions can limit it.

• From digital administration to digital control

This increase in capacity takes on a different meaning when it is integrated with mechanisms of political control. The same technologies that make an administrative service more efficient can also make it easier to identify, monitor, or restrict an individual.

Freedom House’s Freedom on the Net 2025 report estimates that internet freedom has deteriorated for the fifteenth consecutive year. Of the 72 countries covered by the study, the organization finds a deterioration in 28 and an improvement in 17. These results are based on Freedom House’s own methodology and should be interpreted as such, but they document the spread of censorship, surveillance and control practices of the digital information environment.

Digital authoritarianism can be broken down into four functions. The first is to see. Cameras, administrative data, communications, transactions or activities on platforms increase the possibilities of collection. The government does not necessarily own all this information directly: it can be held by companies or distributed among several administrations. Political capacity therefore depends as much on legal and institutional access to data as on its existence.

The second is to identify. Information becomes much more useful when it can be linked to a person or organization. Digital identities, biometrics, facial recognition or unique administrative numbers reduce the ambiguity between activity and identity.

The third is to analyze. Automation makes it possible to search for behaviors, networks, or anomalies in data sets that are too large to be examined manually. Artificial intelligence potentially increases this ability by making it possible to classify texts, images, and behaviors, produce scores, or identify correlations.

The fourth is to act. Information only has a political effect when an institution can draw a consequence from it: filter content, open an investigation, block a transaction, modify access to a service, intervene physically or simply signal to an individual that he or she is being observed.

China is the most studied field of this articulation between information capacity and control. However, the image of a perfectly centralized and omnipotent apparatus must be avoided. On the contrary, field research on Chinese local administrations shows unequal capacities. A study published in The China Quarterly in 2024 describes local offices responsible for controlling cyberspace facing limitations in personnel, technology and authority, offset in particular by the outsourcing of certain surveillance activities, the organized reporting of content and the secondment of personnel.

This observation is important because it corrects two opposing but equally simplifying representations. The digital state is neither a simple neutral tool for administrative efficiency, nor necessarily a perfectly integrated surveillance machine. It is made up of organizations, budgets, competences and hierarchical chains that can fail, compete with each other or have incompatible information.

Data integration reduces some of these problems. Research on local Chinese platforms shows that the pooling of information previously dispersed between bureaucracies can improve the capacity for intra-administrative control and facilitate intervention in the face of social mobilizations. The strategic value of data therefore lies not only in its quantity but in its ability to be linked, interpreted and converted into action.

This logic explains why artificial intelligence is potentially a new step. It does not automatically give the state new information; it can allow it to process more quickly the information it already has. Its political effect will therefore depend largely on the pre-existing information architecture.

• The informational paradox: seeing more does not necessarily mean knowing more

Technocratic authoritarianism contains its most important contradiction here. Technocracy is based on a promise of knowledge: the decision would be better because it is based on experts, data and measuring instruments. Authoritarianism, however, reduces certain mechanisms by which unfavorable information flows back to the decision-maker.

The problem is an old one. In a hierarchical organization, a subordinate evaluated on his or her results has an incentive to present these results favorably. This incentive increases when the sanction of failure is significant and the evaluator depends on the subordinate to obtain the information that allows the failure to be precisely measured.

The literature on China provides a particularly developed illustration of this mechanism. Jeremy Wallace studied the discrepancies between certain provincial growth statistics and less politically visible indicators. His analysis concludes that the most politically sensitive data were more likely to exhibit anomalies during periods of local political transition, which is consistent with the hypothesis of strategic manipulation of statistics.

The problem does not mean that all Chinese statistics are fictitious, nor that the manipulation is unique to China. It reveals a more general difficulty of any performance-based administration: when the indicator becomes a career objective, agents can optimize the indicator rather than the reality it is supposed to measure.

China’s central government is familiar with this problem itself and has developed mechanisms to circumvent bureaucratic filters. Jie Gao describes strategies for central authorities to access raw data more directly in order to bypass local levels that may change information. Pan and Chen also showed the importance of online complaints as a potential source of information about the behavior of local authorities, while documenting how officials may seek to distort information flowing back to the center.

Digital technology can therefore be interpreted as a partial attempt to solve a problem produced by the hierarchy. A national database reduces dependence on the report written by a local official. A sensor directly transmits a measurement. A centralized platform makes it possible to compare several territories. Artificial intelligence can detect inconsistencies that the human administration would not see.

But none of these solutions completely eliminates the problem. The data may be correct but misinterpreted. It can measure what is easily quantifiable rather than what is politically important. Actors can change their behavior once they know the metric used to evaluate them. An algorithmic system can reproduce the biases of the data that feeds it. Information can be available without reaching the appropriate decision-maker. Finally, the center can have accurate information and still choose a wrong policy.

Informational sophistication should therefore not be confused with political rationality.

This distinction is fundamental because it calls into question an implicit assumption of technocratic authoritarianism: the concentration of decision-making would reduce the noise of political competition and allow for more rational action. It can indeed reduce certain coordination costs and speed up execution. But protest, independent media, political oppositions, professional organizations or associations also serve an informational function: they produce signals that the administration would not necessarily generate itself.

Contestability is therefore not only a normative value. It can be analyzed as an institutional technology for detecting error.

This does not mean that democracies always produce better information. They may face polarization, disinformation, organized interests, siloed bureaucracies, and short-term electoral incentives. Rather, the difference lies in the multiplicity of channels through which contradictory information can become publicly visible and politically costly to ignore.

Technocratic authoritarianism tries to solve this difficulty by developing its own channels of feedback: surveys, complaint platforms, local experiments, inspections, statistics, internal polls or analysis of social networks. It thus seeks to obtain the informational benefits of society without necessarily accepting all the political consequences of its autonomous expression.

This tension is probably its most important structural limitation. The more the system wants to govern scientifically, the more it depends on information that it can itself contribute to distorting.

• Is artificial intelligence deepening technocratic authoritarianism?

Artificial intelligence potentially changes this equation because it makes it possible to increase the volume of information that can be processed without proportionally increasing the administrative staff. Its political importance therefore lies not only in the sophistication of certain tools, but in the reduction of the marginal cost of analysis.

China has explicitly integrated this perspective into its national strategy. In August 2025, the State Council adopted the guidelines of the « AI Plus » initiative, which aims to deepen the integration of artificial intelligence into science, industry, consumer affairs, public services, governance and international cooperation. The document sets a horizon for the widespread distribution of intelligent agents and terminals by 2030 and envisages an economy and society that will be profoundly transformed by AI by 2035.

However, it would be simplistic to read this strategy only from the perspective of political control. Official texts place a strong emphasis on productivity, scientific innovation, industry and services. Artificial intelligence is first presented as a general technology of economic and administrative transformation. It is precisely this versatility that gives it political significance: an infrastructure developed to optimise the economy or services can also strengthen the processing capacities available for other state functions.

Saudi Arabia is pursuing a similar logic of integrating AI into economic and administrative transformation. Its Vision 2030 report for 2025 presents the goal of a more data-driven government that uses artificial intelligence to support decision-making. The UAE is also embedding data, proactivity, and automation into its digital government strategy.

AI can produce three distinct transformations. The first is administrative technocratization. Automated systems assist civil servants, process requests, detect anomalies and reduce delays. This evolution can occur in any political regime and is already one of the main fields studied by the OECD.

The second is a technocratization of political decision-making. Models, simulations and predictive tools can give more weight to administrations capable of producing them. The risk is not necessarily that the algorithm decides on its own, but that political choices involving trade-offs are gradually presented as the objective consequence of technical optimization. The debate then shifts from the end to the method: a decision becomes more difficult to contest when it is presented as the result of a model that few actors are able to audit.

The third is an increase in the capacity for individualized control. Automation can make it possible to analyze more content, link more data, and detect certain activities more quickly. Freedom House specifically believes that the choices made around the next generation of artificial intelligence tools will have significant consequences for digital freedom, in an environment where surveillance and censorship are already widely documented.

These three developments are not the necessary steps in the same trajectory. A state can largely automate its administration without developing a comparable system of political control. An authoritarian regime can use surveillance technologies while maintaining a poorly performing bureaucracy. A technologically advanced state can impose strong legal limits on the uses of AI.

The challenge therefore lies less in the technology itself than in its institutional structure.

By 2035, three trajectories can be distinguished without assigning them a determined probability. The first would see AI increase the efficiency of existing administrations. The second would lead to an increasing role for algorithmic expertise in the justification of decisions. The third would combine integrated databases, identification, predictive analysis and the ability to intervene in more individualized control architectures.

The discriminating factor between these trajectories will be less the availability of artificial intelligence models (which tends to spread internationally) than data governance, institutional capacity, access rules and mechanisms to challenge their use.

• Policy implications: Ability to execute and contestability are not necessarily opposed

The study of technocratic authoritarianism finally leads to a question that directly concerns democracies. The comparison is frequently formulated around an opposition between authoritarian capacity for execution and democratic slowness. This opposition is analytically insufficient.

The concentration of power can reduce some decision-making costs. A policy can be launched without lengthy parliamentary negotiations, some procedures can be expedited, and administrative resources can be concentrated around a national priority. But the speed of decision and the quality of the decision are two different variables. A bad policy that is quickly executed does not become better because it is executed effectively.

Conversely, protest can slow down a decision while improving the information available, revealing unanticipated effects and allowing it to be corrected. The strategic question for democracies is therefore not to choose between expertise and politics, but to organize their articulation.

This issue becomes more important with artificial intelligence. The OECD insists on the potential benefits of its use by administrations while highlighting the risks related to data quality, opacity, privacy, representation and accountability. A democratic administration that renounces AI for fear of its authoritarian uses could lose capacity. An administration that adopts it without a control mechanism could, conversely, gradually weaken the safeguards that distinguish its use from that of a system that is not contestable.

The first implication is therefore to treat state capacity as a strategic issue in its own right. Democracies need administrations that can plan, use data, recruit technical skills, and execute policies quickly. Administrative efficiency does not belong to any type of regime.

The second is to preserve technical contestability. An algorithm-assisted decision must be able to be examined, explained and, where its effects affect important rights or interests, challenged. This requires technical expertise in parliaments, courts, independent authorities and oversight organisations. A democracy cannot supervise an algorithmic administration with institutions that do not understand the systems they are to control.

The third concerns data architecture. States will need more interoperability to improve their services, but absolute centralisation is not necessarily the only way to achieve this efficiency. Access rules, traceability of consultations, purpose limitation and control mechanisms can make it possible to exploit information without making all data available to all government functions.

The fourth concerns administrative culture. The informational paradox studied in authoritarian regimes has a less intense version but is also present in democracies: any bureaucracy can produce conformism, manipulated indicators or excessively favourable feedback of information. The protection of internal contradiction, independent evaluation and the ability to publicly acknowledge failure are therefore not only political principles; they contribute to the quality of the information available to govern.

The strategic challenge can finally be formulated in a relatively simple way: to increase the State’s capacity to execute without reducing the company’s ability to contest its action.

This issue is expected to become one of the main institutional challenges of the digital transformation. AI increases the potential capacity of administrations. It does not determine how this capacity will be governed.

Conclusion

Contemporary technocratic authoritarianisms are not an entirely new political regime. Their development is part of older traditions: professional bureaucracies, developmentalist states, planning, administrative meritocracy, government by objectives, and the search for legitimacy through performance. Digital technology and artificial intelligence do not create these mechanisms; they increase their reach.

The interest of the notion therefore lies in the identification of a combination. Some political systems associate a low contestability of power with a strong valuation of administrative competence, measurable performance and informational capacity. China, the Gulf monarchies and the Singaporean case show that this combination can take very different institutional forms. There is no single model or necessary trajectory towards a homogeneous form of technocratic government.

The first transformation concerns legitimation. Competence and results can complement ideology, nationalism, tradition or the personalization of power. A state that can quickly build an infrastructure, digitize its services or effectively manage a crisis can use these results as a justification for its political organization. But this strategy also creates dependency: when a system justifies part of its authority through its performance, their deterioration becomes politically more important.

The second transformation concerns knowledge. Digitalization allows the state to collect and integrate more information, while artificial intelligence increases its processing capabilities. These technologies can improve services, coordination, and decision-making. They can also increase oversight and control. The difference is not primarily in the tool but in the institutions that determine its use.

The third transformation concerns precisely this capacity for control. Data integration, digital identification and automation make a more individualized intervention technically possible. But the existence of this capacity should not lead to the representation of authoritarian states as omniscient systems. On the contrary, research on Chinese administrations shows problems of coordination, unequal local capacities and strategies designed to circumvent the distortions produced by the bureaucracy itself.

It is in this contradiction that the main limitation of technocratic authoritarianism lies. The system seeks to increase the quantity and accuracy of the information it has while restricting some of the autonomous mechanisms capable of producing contradictory information. Indicators can be manipulated, targets circumvented and bad news filtered out. Digital integration can reduce these problems without eliminating them, because all data still has to be produced, selected, interpreted and turned into a decision.

Artificial intelligence will likely deepen this tension rather than solve it. It will allow more information to be processed and more functions automated, but the quality of the decision will continue to depend on the quality of the data, the objectives assigned to the system, and the ability of institutions to detect their own errors.

Technocratic authoritarianism should therefore be understood neither as the replacement of politics by experts, nor as the simple digitization of old forms of coercion. Its specificity lies in the articulation between the concentration of authority, the valorization of competence, legitimization through performance and the increase of the state’s informational capacities.

This articulation produces real administrative power, but it does not suppress politics. Rather, it shifts part of it towards defining objectives, constructing indicators, monitoring data, and selecting those who are allowed to challenge the decision.

The fundamental strategic question is therefore not whether machines will one day allow a power to know a society perfectly. Such a perspective neglects the organizational, informational and political limits that remain behind technology. The question is to determine to what extent the increase in information capacity can alter the balance between knowing, administering and controlling.

For democracies, the answer cannot be to abandon technology or expertise to authoritarian systems. On the contrary, it presupposes building states capable of using data, recruiting skills, anticipating and executing, while maintaining the institutions that allow them to challenge their decisions and reveal their mistakes. From this perspective, capacity and contestability are not two competing models of government. Their articulation could become one of the main determinants of institutional power in the twenty-first century.