The relationship between artificial intelligence and democracy is neither linear nor unidirectional, but rather represents an interweaving of conflicting forces that simultaneously enhance and undermine the foundations of the democratic process.
AI has profoundly altered the public space within which democratic debate takes place. This change goes far beyond the simple issue of disinformation. We are witnessing the fragmentation of the collective informational experience that historically has been a fundamental prerequisite for modern democracies.
When different citizens live in radically different information realities structured by recommendation algorithms, the common ground necessary for democratic dialogue is lost.
Paradoxically, while AI has multiplied the available sources of information, it has also eroded the ability of democratic societies to reach consensus on what constitutes "fact." This epistemic erosion poses a deeper and more insidious challenge than simply spreading fake news.
Democracies face a fundamental dilemma: decisions about AI require extremely specific technical expertise, but to entrust these decisions exclusively to experts is to remove them from the democratic process. Thus an unresolved tension arises between the democratic principle (decisions made by and for citizens) and the need for specialized expertise.
This tension is compounded considering that increasingly complex AI systems are opaque not only to ordinary citizens, but sometimes to the very experts who develop them. How can democratic control work when the tools that are supposed to be governed exceed human comprehensibility?
Political microtargeting through AI has become increasingly sophisticated in contemporary campaigns. AI systems analyze vast datasets about voters to create highly personalized political messages. Just Security
Research shows that personality-targeted, AI-generated political ads are significantly more persuasive than generic content. PubMedNih
The scale and efficiency of AI enable campaigns to generate personalized content for millions of voters simultaneously, making microtargeting far more feasible and cost-effective than ever before. PoliticalmarketerCSET
Recent studies show that AI-enhanced campaign tools can identify psychological vulnerabilities in voters and create messages that exploit these characteristics. OUPAcademicTechInformed
Democratic concerns are substantial:
Voting systems and election infrastructure are also affected by AI, in both positive and negative ways:
AI offers both opportunities and challenges for democratic participation:
AI is not simply changing the modes of democratic debate; it is profoundly restructuring power relations within societies. Control of AI infrastructure now represents a form of power comparable to, if not greater than, that traditionally exercised by democratic institutions such as parliaments.
We are witnessing a shift in decision-making power from public institutions subject to democratic control to private entities operating according to different logics. This migration of power often occurs invisibly, through the gradual delegation of decisions to automated systems that operate according to parameters that are not always transparent or democratically established.
AI is transforming the very concept of democratic participation. On the one hand, it offers tools for more direct and participatory forms of democracy; on the other, it introduces new cognitive and technological barriers to access. In this process, the value of individual opinion is also being redefined: in a context of increasing algorithmic personalization, how to distinguish between authentically personal preferences and those induced by recommendation systems?
This ambivalence also manifests itself in AI-assisted deliberation systems: they can make complex decision-making processes more accessible, but they also run the risk of oversimplifying issues that would require more public reflection.
Traditional democratic institutions, conceived in a pre-digital age, struggle to adapt to the speed imposed by technological innovation. This temporal asynchrony between the rapid evolution of AI and the slower pace of democratic processes creates a governance vacuum that risks being filled by undemocratic decision-making mechanisms.
The challenge is not simply to regulate AI through existing institutions, but to rethink these institutions to make them appropriate for an era in which artificial intelligence is becoming increasingly central to social, economic, and political processes.
To meet these challenges, a new conception of democratic citizenship that incorporates an awareness of the role of AI must be developed. This implies overcoming both the naive techno-optimism that sees AI only as an opportunity for democratic improvement and the pessimism that sees it only as a threat.
Instead, it requires the development of civic literacy that includes the ability to critically evaluate the influence of algorithms, to consciously participate in debates on complex technological issues, and to demand transparency and accountability from those who develop and implement AI systems with significant social impact.
In the final analysis, the relationship between AI and democracy is not predetermined by the technology itself, but will depend on our collective ability to imagine and build institutions, norms and practices that allow technological development to be directed toward strengthening rather than eroding basic democratic values.
AI can enhance democratic participation by making government information more accessible, enabling multilingual participation, analyzing large-scale public input, and personalizing civic engagement experiences. IeeeOECDevents For example, AI-enhanced translation services can enable language minorities to participate more fully in democratic processes, ECNL while data analytics tools can help governments identify and address inequalities in public service delivery. Nextcity + 2
The most significant risks include: the proliferation of persuasive misinformation and deepfakes that undermine shared facts; manipulation through microtargeted political content; OUPAcademicMediaengagement algorithmic biases that exclude certain groups from democratic processes; and security vulnerabilities in election infrastructure. Sage Journals + 5 Research shows that these risks are not merely theoretical-studies document the persuasive power of AI-generated political content customized to individual psychological profiles. PubMed + 2
Effective regulatory approaches include: mandatory disclosure requirements for AI-generated political content; limits on the types of personal data that can be used for political microtargeting; independent oversight mechanisms to monitor campaign AI systems; and public education initiatives to increase voter awareness of AI-enhanced persuasion techniques. PBS + 4 The EU AI Act provides an example of a risk-based regulatory framework that specifically addresses the use of AI in democratic processes. Europe + 2
Citizens should have meaningful opportunities to shape AI governance through participatory mechanisms such as citizen assemblies, public consultations, and ongoing stakeholder engagement processes. Brookings Research shows that engaging diverse stakeholders throughout the AI life cycle leads to more reliable systems that better reflect societal values. Adalovelaceinstitute + 13 Successful models such as the Camden Data Charter demonstrate how citizen participation can establish ethical frameworks for the use of AI in public services. Oecd
Protection strategies include: implementing robust cybersecurity measures such as multi-factor authentication; training election officials to recognize AI-enhanced phishing attempts; developing systems to identify and counter AI-generated misinformation about voting; Sophos News establishing verification processes for official election communications; and creating redundant systems for critical election infrastructure. ABC News + 2 The Cybersecurity and Infrastructure Security Agency (CISA) provides specific guidelines for election officials on mitigating AI risks. CisaCisa
AI could transform citizen-government relations by enabling more personalized public services, creating new channels for civic participation, Nextcity automating certain government functions, and potentially shifting power dynamics in democratic systems. EffOECD The direction of this transformation depends largely on governance choices-whether AI is implemented in ways that enhance democratic accountability or concentrate power in technical systems with limited oversight. ScienceDirect + 2
International cooperation is essential to establish shared standards, prevent regulatory fragmentation, address cross-border impacts of AI, and promote alignment around democratic values in AI development. Oecd + 2 The OECD AI Principles represent a model for international coordination, providing a common framework adopted by 47 jurisdictions globally while allowing flexibility for national implementation. OecdBrookings
Ensuring that AI benefits democracy requires: proactive value-based governance frameworks; meaningful transparency and explainability requirements; participatory approaches to AI development and regulation; investment in digital literacy; protection of fundamental rights in AI applications; and mechanisms to hold AI systems and their developers accountable. Eff + 4
Evidence suggests that anticipatory governance approaches are more effective than reactive regulation. OecdBrookings
AI presents both significant opportunities and challenges for global democratic systems. Our comprehensive review suggests several key conclusions:
.svg)
.svg)
.svg)
.jpeg)
.jpeg)