In this paper we will contend that the use of artificial intelligence (AI) for political purposes, as well as in other areas, is having a detrimental efect on the collective identity of Western liberal democracies. To explore this idea, we will analyze three key concepts in classic political theory- those of sovereignty, individual rights, and the rule of law- which we contend are in urgent need of reviewing in light of the new dynamics AI is creating. We also underscore what we understand to be a precedent in the erosion of the principles of liberal democracy in the early 20th century: militant democracy. Our conclusion is that we must aim our eforts at rethinking liberal democracies in a way that allows for the assimilation of the technological progress as part of our political process without compromising its core values in doing so. The concept of Linked Democracy can serve as a starting point for that discussion.
Since the late 20th century, there has been an exponential increase in the development of
technological/digital possibilities and of artificial intelligence (AI) specifically: “digital identity,
the web of data, the internet of things, big data gathering and analysis, management algorithms,
semantic representation of languages, use of intelligent decision-making programs, artificial
robot societies, models of distributed multiagent coordination systems and the unequal
distribution of information” [
In the following pages we will highlight some questions liberal democracies have had to face historically, but to which we have still not been able to respond to adequately: those, we suggest, are the questions regarding the shortcomings of democracy in the face of internal threats to its system.2 The 20th century’s rise of totalitarian ideologies kindled a debate over the legitimacy of “militant democracy”, which is an interpretation of democracy that allows for antiliberal measures to be put in place by a democratic government in order to manage the influx of antiliberal ideologies. In our present day and age, we will contend, we face a situation with a similar underlying question, i.e., how should democratic actors handle the use of the digital tools that may render democracy vulnerable if gone unchecked.
We will begin by exposing the fact that the dilemmas we face today regarding AI in our political systems are not entirely new but part of a bigger problem that was already brought to our attention during the 20th century’s clashes between democratic regimes and totalitarian regimes. Even though the threat was then a direct, open attack on the liberal and democratic principles while AI appears to be ideologically neutral, we will argue that there might be a significant parallel between the political efects of a totalitarian regime and the efects, albeit in a more subtle (and perhaps dangerous) way, of a nominally democratic regime pervasively run by autonomous technology.
We then move on to explore three alternative approaches to this question each of which point at a somewhat problematic issue that springs from this phenomenon. These “red-flags” requiring attention are in our estimation: (i) what powers are the people sovereigns of, (ii) the individual rights legitimation for government and (iii) the rule of law as the only valid instantiation of a democratic regime.
Afterwards we will attempt to clarify in what way in a democracy the implementation of certain digital features can make us reconsider the expectations we have for our regimes and our governments. The open-ended possibilities of AI paired with its unabridged use by those in power can lead us to question our collective identity as a political community and to question the unifying principles behind this community. We shall consider if the (re)definition of the identity of liberal democracies is being played out in real time without the people having an adequate place or space to deliberate it.
Lastly, we will summarize our conclusions.
The concept of Militant Democracy was introduced into comparative constitutional law and
political science by Karl Lowenstein, a German émigré to the US, in a series of articles written
in the 1930s and 1940s. In them, he reflected on the dificult task of tackling the rise of fascism.
2Jan Werner Müller distinguishes between external threats to democracy, which are historical (fascism, communism,
and other forms of totalitarianisms) and internal threats, which are more recent and come from within the democratic
system, such as extremist religions (that may even aspire to a religious-political power) or domestic terrorism
of various natures. See J-W. Müller, A “practical dilemma which philosophy alone cannot resolve”? Rethinking
militant democracy: an introduction, Constellations, 19.4 (2012) 536-39
For him, the only way for democratic states to “withstand the skilful exploitation of democratic
rights to subvert democracy from within in was to abandon what he took to be an “outdated”
view of liberal democracy according to which all voices should be accorded free expression
and participation” [
In short, militant democracy was all about applying the normative resources it is legitimate
for a democracy to use to defend itself from authoritarianisms and totalitarianisms. The three
main strategies being: (i) to concentrate power in the executive (ii) to use emergency powers
(iii) to pass ad hoc legislation to restrict rights of expression, participation, and assembly. [
It is easy enough to sympathize with the premise: democracy should be legitimized to use illiberal tools in the face of potential domestic, as well as foreign, anti-democracy threats. From the perspective of democratic theory, however, is this consistent? And pragmatically speaking, is there evidence of it working? There was never enough serious attention given to the problems addressed by militant democracy during the 1900s, and the topic was never therefore satisfactorily dealt with -let alone resolved. The most likely reasons being a diferent stance in the interpretation of democratic theory, on the one hand. In the past century, authors would take democracy as a whole, neglecting the internal institutional variety of the regimes. On the other hand, the empirical focus was put on political economy and social policy, not the constitutional institutions:
Recent comparative constitutional law and political science literature has converged
on the principle that democracies have a right to defend themselves against their
enemies-even in the absence of violence-. To accomplish this purpose, democratic
states can enact and apply formal rules restricting expression and participation,
subject to impartial oversight of their application. Beyond these basic principles,
little convergence emerges in the practice of contemporary democracies.” [
On another note, despite the end-of-war peace, totalitarian communism had been preserved alive and well at the very heart of Europe. The Cold War too would serve as a permanent reminder of the anti-liberal/ anti-capitalist threat to the democratic world order. In the face of those events, doubling down on the rule of law and on individual rights seemed like the right thing to do. However, and despite these actions, the big questions that the 20th century forcefully raised had remained essentially undealt with: how does a democracy secure for itself the necessary tools to channel threats to its system in an efective way without betraying its principles? There were limits to the efectiveness of the rule of law and to the eficacy of the democratic process when it came to securing the continuity of the system. And the solution had been to reinforce those. However, the specific question of what to do if and when their frailty became flagrantly apparent again was not resolved.
The question had been raised and had been dealt with to some extent, -both in theory, as
we have mentioned in the case of militant democracy, and in practice, in how democracies
strategized and fought the war3 -, and some grapple with it still today [
Be that as it may, nowadays, with the advent AI we find ourselves in a complex, contemporary
version of a similar updated issue: how should a democracy handle the technology that has the
potential to disrupt the democratic process in a way that poses a threat to its own continuity?
Our contention is that it is not a matter of regulating via the tools we currently have; “what needs
to change is not the content of the regulations, but their form, the very notion of regulating”
[
We contend therefore, that anything that meddles digitally or technologically with that natural democratic process -as does AI-, is a threat to democracy in a parallel way that totalitarianisms are a threat by meddling with the process ideologically or institutionally. To further clarify how it is that that can occur, specifically in the case of new technologies, we need to be granular about what the building blocks of democracy are and analyse one by one where and how a key aspect of democracy is being subverted. This may allow us room to come up with a more nuanced interventionism than goes beyond simply over-regulating. 3We still debate to this day how to justify without blushing that the dropping an atomic bomb on innocent civilians is a legitimate democratic course of action.
“As in other disruptive activities, there are things in AI that we can do, some things that we
should do, and more significantly there are also things that we should not do even if we could.
The problem is that, for now, it is not clear where the red line should be drawn.” [
The urgency to discuss this topic comes from our day-to-day reality. Current digital developments have been increasingly granting democratic governments uncurtailed access to standard, as well as custom-designed tools for surveilling and controlling citizens (facial recognition, registry of infractors’ identities, digital profiling, data re-identification…). And what is more, it allows them the means to easily do so without the afected citizens’ knowledge, let alone their consent. The justification for such actions is once again (at least in the West) to further guarantee the security of the rights of their citizens (for instance, when the Canadian PM ordered banks to freeze the accounts linked to trucker protests over a Covid vaccination mandate “with no need for court orders”4). Therefore, the core issue at hand is still the previously identified unresolved issue “how antidemocratic can a democracy get in the name of democracy?” That happens to be the precise question we need answered in order to establish the uses, extent, limits, ethics of these digital and AI processes in a democratic society. We keep unsuccessfully trying to regulate the use and implementation of these tools but consistently get stuck at the point where we realize that the potential regulators of the tools are the ones who have free access to them. More on this later.
Another reason why we do not give a clear answer to these questions could be because we
do not usually frame the correct question for it. We tend to diversify it and break the problems
into many diferent questions depending on the area of concern, which makes for a heap of
potential partial-answers: legal, financial, image rights, privacy … We here set out to focus on
a core problem at the heart of all these issues: what is the identity of democracy in the 21st
century? I.e., what do we as a liberal society stand for, and what are we not willing to tolerate,
even under the guise of security-guarantees? We are aware of militant democracy justifying
anti-democratic measures under certain circumstances to protect democracy. Invernizzi and
Zuckerman, however, contend that “if democracy is to be understood as a form of government
based on the principle of freedom as a collective self-government, this suggests that it must
inevitably be willing to assume a certain measure of political risk.” [
Ever since the 1980’s, the expansion of algorithmically controlled technology has been exponentially furnishing democratic (as well as un-democratic) governments with tools for an illiberal surveilling and control of citizens without their knowledge or consent. Is this a form of militant democracy? Or is it a form of authoritarianism? Our argument here is not that all democratic governments are secretly evil undercover totalitarianisms banking on citizens’ ignorance and passivity to overturn democracy in a 21st century version of Orwell’s Big Brother. Not at all. Our contention is simply that as things stand today, politically and technologically, it is not far-fetched to alert that we may be headed down a path leading to something akin to what Yuval Noah Harari calls digital dictatorships. Whether we do it willingly or not is one question, whether knowingly or not, is quite another. These tools do indeed have the power to become dictatorships in that they contain control, surveillance, oppressive, uniformity-inducing potential while at the same time they have the ability to obstruct the free flow of information and knowledge that thwarts a natural democratic process. This is the conundrum we must face.
This reality opens up a scenario that we need to address the sooner the better. For starters, we must discuss if we justify using AI for political purposes as a guarantor of democratic rights or if we see it as an undesirable expression of militant democracy which could backfire on us in unforeseeable ways. We also must discuss if given the potential of AI and its elusive nature, we consider it to be one same element to be contended with or if we are to consider each device separately as an independent tool to be dealt with within its own area of application and following the standards and protocols pertaining to each discipline. We will open the possibility of not considering AI services as tools but as powers in themselves. This idea implies rethinking the theory of popular sovereignty that undergirds democracy. We will also suggest reconsidering the possibilities that the classical rule of law theory ofers us in this contemporary scenario and will discuss the need for broadening the possibilities of administration of a liberal and democratic society, such as through linked democracy. Lastly, the traditional notion of individual rights as being secured by democratic governments, and thus legitimizing them, comes under attack with AI having the autonomy and the potential to do away with them overnight, unless actively avoided by democratic agents. We have called all these “red flags”; areas of fundamental importance to democratic theory, as relevant today as the day they were developed, but whose endurance is being put to the test and whose resilience is being taken for granted.
We realize that the traditional doctrine of the rule of law and of individual rights doesn’t make the cut for our present situation with AI. We have not developed standardized tools to address it yet and are at a loss as to which is the appropriate way to approach its development and impact -that is, one consistent with the democratic-liberal principles. One of the dificulties lies in that we lack clarity as to the target to tackle when it comes to AI. What is the goal when it comes to approach this disruption? To engage in the protection of citizens’ rights at any cost? To reinforce the rule of law and extend it to include AI? Should we accept AI as a new reality and simply adapt the diferent aspects of society to the new technological-digital possibilities? To formally articulate AI into our current systems in order to generate in the citizens accurate expectations regarding certain behaviours and sanctions?
What follows are some reflections on a few topics we consider useful to address and which might help us begin to elucidate what our priorities in this regard should be. In its classical modern liberal sense, democratic regimes exist through instruments such as the rule of law to ensure that the locus of power remains in the individuals and to protect the citizens’ individual rights above all else. This basic premise could be a starting point to explore if and how AI afects the foundations of our regimes. Let us go by parts.
Despite states’ executive authority, it is a constituent element of representative democracies
that the power resides in the people. The concept of popular sovereignty entails that it is the
citizens’ prerogative to decide how that power is meant to be articulated and what its limitations
should be. However, could it be possible that recently a new source of power has risen out of
the blue and dodged this filter? An element of the technology management challenge facing us
could be that we have been misrepresenting AI in our debates over how to best handle it. We
may have been trying to accommodate AI within our current legal structures when, in fact, the
tools for algorithmically controlling citizens’ privacy and the access to and dissemination of
information may not be simply tools for power after all; they may constitute a form of power in
and of themselves. The case could be made that nowadays some new powers have emerged
given the fact that “whomever controls the relation between data and metadata of the elements
of a system is able to construe its power of implementation, that is, the extent of the ecosystem
of its utilization.” [
This becomes problematic in another are as well since this form of power would not have
been equally distributed: “This is an asymmetric relationship: citizens, consumers, users, may
be totally identified and the scenarios and context may be delineated, yet they may never have
access to the general knowledge that the system and its administrators are privy to” [
We here are focusing on the issue of the locus of power, and on the fact that these are very aggressive tools for the exercise of power and that are now being used freely, without the legitimation or even scrutiny of the public. This is especially serious since in most cases the use of these tools incurs in incompatibilities with the liberal principles of the democratic regime in ways that go way beyond this lack of citizen supervision. It is thus worth looking into if it is a wise and necessary course of action to attempt to reclaim this digital power from the government and return it to the people. After all, “we the people” have not decided how we want our governments to use this kind of tech and AI -or if they should even use it at all- let alone expressed this decision in a procedurally valid way.
As for surrendering this possible new power to the people, it appears this
deliberative/epistemic democratic process has not taken place probably, and most importantly, because there
doesn’t seem to be a designated arena to do so. We are indeed in an urgent need for one. The
model of linked democracy is a good example of how we can begin the integration of AI and
new technologies into the normal political process with all their power, but at the same time
prevent them from taking over -or being warped by- classical political/legal dynamics that prove
to fall short in this new scenario [
Modern liberal democracy was originally devised as a mechanism for establishing a government
by consensus whose legitimacy hinged on being the ultimate guarantor of individual rights in the
face of absolutism. Since then, a complex evolution of both the concept of rights and the content
of those rights have placed this classical liberal legitimation for government in a very delicate
position. “The gradual acknowledgement in continental Europe of civil rights, political rights
and, finally, the ‘social rights’ has been matched by a gradually more selective, legally imperfect,
and politically reversible guarantee of rights. A sort of ‘law of decreasing efectiveness’ as to
the protection of individual rights may be argued. Such ‘law’ is due to the diferent relationship,
which has gradually been established in Europe between the acknowledgement of rights, on
the one hand, and the functional requirements of a political system correlated with the market
economy, on the other.” [
In this context, when AI irrupts into the political scene ofering the powers that be the opportunity of accessing privileged ways of efectively protecting rights at a scale never before imaginable (like those earlier mentioned), it seems sensible that they would jump at the opportunity. After all, our governments are still the legitimate guarantee-mechanism for all the sorts of modern rights. Nevertheless, it’s worth considering that we may have collectively fell prey to the belief that since AI can help us in that mission, morally, it must. Even if it requires violating certain individual rights to better protect others, the general feeling may be that a government can’t responsibly pass up the opportunity to increase its power to this end. Therefore, in this context, what we suggest questioning is not quite the same as it is with militant democracy, i.e., “is it legitimate for our state to protect our rights in non-democratic/liberal ways?” Whether it is or whether it is not, we know for a fact that they already do. Governments (democratic or otherwise) have, and are, using ethically questionable technological means in their management of citizens and their institutions.5 The question revolves not so much around if the state is serving its purpose to defend individual rights, we know that is what it purports to do; the question should be centred around to what degree is it violating that very purpose, while allegedly pursuing it? In attempting to uncover if our rights are being violated by government at all, we reach the real crux: when has our democracy signed of on that to any degree? In other words, the red flag is, in this case, the awareness that something must be of with the democratic process at some level when we have already been able to seamlessly assimilate these algorithms into citizens’ everyday lives in a way completely unbeknownst to them -despite the profound consequences for their political, social, financial, and private lives. It begs the question: Is the state still really out to protect our rights at all anymore?
There is a lot to be said in favour of technology aiding in the governments’ functions. It ofers a degree of efectiveness, speed, sustainability, reliability, consistency… all at once, that is in fact unparalleled by any other means even known. The list expands in surprising new ways every single day. And so, since the digital tools of power may help us with protection of individual rights, should they therefore necessarily do so? We suggest the following question be used as criteria for answering the previous one: Is it legitimate from a standpoint of democratic theory that it do so? We can judge that by considering the following issues: (i) the first question is to deliberate if the use of a specific AI technology in a specific instance unequivocally implies a violation of some individual right. Secondly, if the answer to the first question is “yes”, we decide (ii) whether we legitimize violating some rights, some times, if necessary, in order to defend the system of rights’ overall integrity. Lastly, if the answer to the previous consideration is “no” (i.e., the ends do not justify the means), then (iii) could those digital mechanisms be reversed by now? Or more interestingly (and realistically), given this state of afairs, how can this process be channelled in a way that will comply with the democratic spirit?
Although we cannot go into all these necessary questions here, it is worth noting that once again we may find that the linked democracy model, we discussed earlier can prove to be a helpful tool for approaching this topic in a theoretic/practical way. The linked democracy model integrates the technology into the democratic process itself and thus is able to creatively link the source of the potential threat with the very people whose rights it potentially threatens, therefore, with those who should be making the actual decisions and deliberations regarding it. Moreover, what linked democracy also does collaterally, is make us rethink the classical structures that have held democracy in place so far and makes us wonder how they fit into our contemporary digitally driven world.
The classical rule of law that endures still today is an example of these structures that have been
inseparably associated with modern democracy since it beginning; it stands to reason, since the
principles of the rule of law are joined at the hip with the democratic doctrine of individual
rights [
We have mentioned before that governments have already, resorted to implementing AI on their citizens, accessing and implementing a new form of power, yet to be legitimately claimed and contested. It is not foolish to consider the possibility that in the face of its gradually being rendered obsolete, it is the rule of law system per se, and not the protection of individuals’ rights, that is benefiting from these tools and the political, legal, and ethical limbo they are in right now. Could it be that the rule of law system itself uses the unabridged access to this novel power as a short-cut that helps to “compensate” for its limitations in the face of its current challenges? In other words, does democracy absolutely require the government using these digital tools for the protection of citizens’ rights? Perhaps they do, or perhaps not, but in any case, it is clear that if we agree that they in fact constitute a form of power, AI tools need to be submitted to the democratic process first and foremost, regardless of the ends they are purported to be pursuing. The goal is to ensure that the power is located where it belongs in democracy (in the people); that the priorities of power are the democratically legitimate ones (protecting individual rights); and that the legal and political structure that sustains the democratic process is the optimal one; and that may require a revision of the rule of law. It is possible that we may have been mistakenly conflating a democratic system with the rule of law system, as in two sides of a same coin. This problematic identity may have led democracies to make some unnecessary concessions when it comes to citizens’ individual rights due to a lack of alternatives, not out of democratic conviction, since democracy does not inescapably require the rule of law.
An exception to the rule of law as sole and ultimate guarantor of individual rights is also
found in the British tradition. Edmund Burke was “appalled by the notion that rights should
be the normative building block of politics (…) he saw both rights and obligations as rooted
in the inherent traditions that give life to the political communities. This view of the primacy
of collective traditions over individual rights and obligations is characteristic of the world
view we often call ‘communitarian’” [
Yuval Noah Harari [
It is essential to be aware of the workings of democracy being hijacked from within itself, since it is the procedure precisely -the democratic procedure- that produces political outcomes one may disagree with, but can nevertheless stand by and defend as legitimate regardless of one’s diferences, political or otherwise. That is the very miracle of democracy: the way in which it has the ability to turn the political process into the standard for success and how that very process knits together all those involved in it -the deeper goal- and not only benefits those whose ideas are actually implemented. If we meddle with the democratic procedure, either because we unnaturally manipulate it or because we artificially subvert its natural flow, what we do is we erode its legitimacy and eventually question the validity of a structure that stems from manipulated citizen interactions and decisions. Therefore, it is important not to demonize the technology or the digital resources themselves, but in turn to focus our attention on the algorithms, on the opaque manipulative decisions that are being made for us, without our consent or knowledge; in the best case, in the name of protecting us, in the worse case …we don’t want to know.
This degree of digital-technological AI possibilities specifically targets Isaiah Berlin’s idea of
liberal value pluralism [
Today’s democracy’s problem is not one of internal threats to the system or want of
governmental resources (both justification for militant democracy and illiberal use of AI, respectively).
If anything, and issue that is hindering their development is too much homogeneity, too much
control and self-censorship. Tech is disruptive, among the other issues we have discussed
already, because it forces us to reconsider the value of pluralism and its role in our social and
political environments. We contend that the need to underscore this key instrumental role
of pluralism in securing a healthy democracy is a very good thing, and it cannot be stressed
enough. Pluralism as a value from the perspective of each individual’s right to self-expression
and self-assertion also deserves mentioning, but it is its role as a procedural lubricant that is
endangered by the uniformity machines tend towards by nature. That is not to say technology
cannot ofer perks to the whole of the democratic process. Ober underscores the element of
Athens’ democracy’s success based on “the simultaneous innovation-promoting and
learningbased context of democratic institutions and culture” [
Since the late 20th century, we have experienced an exponential increase in the development of artificial intelligence and digital and technological possibilities. These have impacted citizens’ ordinary lives and have also introduced a profound distortion of political and social power relations. However, we have not yet addressed the uncomfortable changes that these shifts are producing in the political identity of liberal democracies in the West and the impact they have on the very nature of governmental power.
In these pages we highlighted the question liberal democracies have yet to respond to: the question regarding the limitations of democracy in the face of internal threats to its system. Militant democracy, as we discussed, is an interpretation of democracy that allows a democratic government antiliberal measures in order to manage the influx of antiliberal ideologies. Nowadays, the question we are presented with is surprisingly similar: how should democratic actors handle the use of digital tools that may produce a questionable outcome for democracy. While AI appears to be ideologically neutral, we argued that there might be a parallel between the political efects of a totalitarian regime and the efects, albeit in a more subtle way, of a democratic regime pervasively run by autonomous machines.
We explored three alternative approaches to this issue with the goal of analysing some key problematic issues that spring from this phenomenon. The first one was what powers are the people sovereign of, and we realised AI may be a new and emerging power on its own. The second one had to do with the individual rights legitimation for government, and we questioned whether that could still be considered the main foundation given the state of modern rights and the efect on them that AI is producing. Lastly, we questioned the rule of law being the only valid instantiation of a democratic regime, particularly in the digital context we find ourselves in.
The open-ended possibilities of AI paired with its unabridged use by those in power can lead us to question our collective identity as a political community and to question the unifying principles behind this community. We considered if the identity of liberal democracies is not only being played out in real time but that this is happening without the people having a place or space to deliberate it. Linked democracy is a possible pathway to explore in order to move forward democratically instead of backward in the name of democracy.