Abstract
Is it time for politics and political theory to face the challenge of artificial intelligence (AI)? It seems to be the case that political theory constantly lags behind technological developments. With rapid developments in the field of AI, a common estimate is that technological singularity will probably happen in the next 50 to 200 years. Even regardless of the time frame, the very possibility of superhumanly smart AIs poses serious political questions and calls for some serious political decisions. Luckily, some scenarios are already provided by science fiction. The most common trope is that of a hostile AI taking over and destroying/enslaving humanity. Yet there is another path—AI takes over and makes itself a ruler (openly or behind the scenes), but rules in the genuine best interest of humanity. Such a turn of events, as described in Neal Asher’s Polity novels, may be an improvement compared to present political arrangements. It will, however, probably rob us of politics as such.
Introduction
“Politicians should read science fiction,” A. C. Clarke famously stated (cited in Agel, 1970). This statement, one could argue, should be amended to include political theory scholars as well. Ever since Frankenstein, the fear of uncontrolled, and, moreover, uncontrollable technology has steadily pervaded popular culture, and from time to time spilled over to academic discourse on the future of humans, politics, and society.
This theme of autonomous, out-of-control technology in political thought was comprehensively elaborated in the seminal work of Langdon Winner (1977). However, his insight that “technology is itself a political phenomenon,” and therefore a legitimate object of study for political theorists, was not widely recognized. When discussing technology, most authors in the field still do it within paradigms of technological determinism and instrumentalism, and the constructivist approach, dominant in the field of science, technology, and society studies, is, for the most part, ignored.
Moreover, political theory has, so far, mostly reacted to technological developments, sometimes when new technologies seem to be “within reach,” but more often after they had already been introduced. Catastrophic events were usually the trigger that sparked first public and then theoretical interest. 1 Given the ever-accelerating rate of technological change, this reactive stance, I believe, should change. Political theory should start exploring how its key concepts will be affected when confronted with technologies yet to come. In this respect, the imagination of science fiction authors is a valuable and readily available, but thus far underutilized resource.
All of the above is particularly true with regard to artificial intelligence (AI). With research well under way, inception of the general AI does not seem to be very far in the future. Possible consequences, already extensively discussed in popular culture, should be seriously analyzed, or it may be too late.
In following sections, I shall briefly present definitions of AI, the current state of research and tropes about AIs in popular culture; I will draw an outline of Neal Asher’s Polity universe; I will also discuss some political questions that arise around the construction and use of AIs and finally explore how the existence of AIs, as described by Asher, relates to concepts of technocracy and democracy as well as could rule by AIs be preferable to existing political arrangements, especially to modern democracy.
AIs Are Coming
Research of AI started with McCulloch’s and Pitts’ mathematical model of neurons in 1943 and Turing’s work in 1950s (Warwick, 2012). In those first decades, the main goal was to build a machine that could successfully emulate human behavior. But even then, Turing (1950) wondered if the way machines think could be very different from the way humans do. This line of thought led to a paradigm shift in 1980s, when a new approach to AI research abandoned the idea of copying human intelligence and instead started to explore possibilities of AI “intelligent in its own way.” Thus, AIs could not only mimic, but surpass the human brain, and become “bigger, faster and better” (Warwick, 2012). Today, it seems obvious that certain programs, such as chatterbots or social bots, are quite capable of emulating human behavior, or at least communicate in a manner that is indistinguishable from human communication. Whether this makes them intelligent is another matter altogether. Although they may pass the Turing (1950) test as it was originally envisioned, there seems to be a consensus that the test itself should be revised (Gomes, 2015).
Basically, the very definition of AI 2 has moved from “a machine that is capable of humanlike behavior, at least in those capacities attributed to the mind” (Buchla, 2008) to “systems which match or exceed the [intelligence] of humans in virtually all domains of interest” (Shulman & Bostrom, 2012, cited in Muehlhauser & Salamon, 2012). This goes hand in hand with the notion that once we construct an AI that is as smart as humans, construction of ever smarter AIs will follow, resulting in an “intelligence explosion” or “technological singularity,” a point beyond which every prediction becomes impossible. 3 While predictions about the timeline of AI development vary significantly, the most common estimate is that the first AIs are to be constructed in 15 to 25 years (Armstrong & Sotala, 2015; Muehlhauser & Salamon, 2012), and the technological singularity may be waiting to happen in the next 50 to 200 years. All of this, of course, pertains to “general AI.” Expert systems, also called “narrow AIs” are already being developed and used in a variety of contexts—from stock market trading to medicine and entertainment industry. So it may be a good time for political theory to consider possible issues and outcomes. Whatever the conclusions may be, at least the humanity would not be caught unaware.
AIs in Popular Culture
The notion of artificial intelligences “taking over” from humanity is one of the most repeatedly used tropes in science fiction. In literature, movies and games, sentient machines are usually depicted as villains. Shortly after they become self-aware, they try to enslave or erase humanity, seeing it as either a rival or vermin.
Long before the science fiction genre stood on its own, this was one of the central points of Samuel Butler’s Erewhon. At the beginning of the 20th century, he writes:
There is no security [ . . . ] against the ultimate development of mechanical consciousness, in the fact of machines possessing little consciousness now. [ . . . ] Is it not safer to nip the mischief in the bud and to forbid them further progress? (Butler, 1910)
If the development of machines is to follow the path analogous to evolution of human consciousness, and their behavior toward “lesser beings” is to mirror that of humans, humanity is, in the long run, doomed to become no more than a race of pets or slaves to the machines. Even more terrifying is the possibility that this descent to slavery would be slow and imperceptible, and that the new slave-owners would be kind and benevolent, thus ensuring that “the revolution will not occur in our time, nor hardly in ten thousand years, or ten times that” (Butler, 1910). Following this line of thinking, people of Erewhon decide that the only solution is a preemptive strike—destruction of all machines invented in the previous 271 years. It is worth noting that this was a political decision par excellence, achieved through debate, bargaining, and negotiations, and followed by “reactionary civil wars.”
This trope has had many iterations since. As the 20th century was nearing its end, the threat of renegade AIs became commonplace in popular culture. Skynet from Terminator (1984) soon became synonymous with an out-of-control, hostile AI.
There is, however, another take on this trope. Some authors agree with the assumption that sentient machines/AIs will take over from humanity. But, in their view, AIs would not act as tyrants or slave owners, but rather as shepherds or benevolent kings, ruling in the best interest of humanity. Most prominent example are found in Asimov’s work. Ruling intelligent machines feature in many of his “robot” stories, and such an arrangement usually turns for the best. Asimov furthered the notion that humanity is not to be trusted with its own survival in Robots and Empire (1986), where he formulates the “Zeroth Law of Robotics,” an extrapolation of the First Law, 4 stating that “A robot may not harm humanity, or through inaction, allow humanity to come to harm.” Following this basic rule, humanoid robots, led by R. Daneel Olivaw, work to protect humanity as a whole, mostly behind the scenes, but not shying away from direct involvement in politics when necessity calls for it. 5
More recently, one of the most elaborate visions of a world (or, more precisely, worlds) ruled by AIs was created by Neal Asher. Some of the issues he tackles in his Polity universe are quite relevant for the political theory of today and tomorrow.
Neal Asher’s Polity
Neal Asher’s Polity universe encompasses 13 novels, 6 but most of the main ideas of concern for this article are explicitly expressed in 5 of them, known as The Cormac Series. Set in 25th century, these novels are packed with fast action and complicated technologies, and Asher does not spend many words describing political, economic, and social arrangements. Information about those aspects of Polity are usually presented in excerpts from publications at the beginning of every chapter, or casually mentioned in the dialogue. In his writing Asher does ask some important, philosophical questions—about the nature of life and death, the essence of personal identity, and humanity. He does not, however, provide clear answers. His views on politics and religion, on the other hand, are very clear and directly stated. They will be discussed in more detail in the following pages.
Polity is a political entity consisting of planets inhabited by humans and ruled by a hierarchy of AIs, supreme among them being Earth Central. It encompasses about 100 worlds in one corner of the Milky Way. Humankind had populated the Solar system, after which expansion continued, in waves, as space propulsion technologies advanced, every 50 to 100 years. Polity’s Line—border consisting of worlds yet to be subsumed—is still shifting in the 25th century, not always further into space.
Economy
Even before they overthrew human governments, AIs were in charge of the economy, working for national governments and international corporations. In the process, they managed to establish themselves as corporate entities, amass significant assets, and eventually hijack the entire economy. While this was not by far the only factor in the rise of AIs to power, it does confirm the conventional wisdom that political and economic power are always intertwined. Having been instated as rulers, AIs continue to control the economy of the Polity, using complex formulas to limit the rate of expansion and resource demand, as well as profit margins. The latter is the source of discontent, since some still wish to be free to exploit the workforce and natural resources in order to endlessly increase their wealth.
Technological breakthroughs have led to the unprecedented abundance, and “everyone has their share of plenty.” While those shares are by no means equal, everyone is provided for and poverty is eradicated. Strangely, cash still exists, being one of the many relics of the past whose resistance Asher ascribes to human nature.
Religion
Asher’s stance on religion is quite clear: He is openly against it, and his views seem to be influenced mainly by Richard Dawkins. Sometimes, for example, in The Line of Polity, this antireligious stance is elaborated to the point of preaching. The plot of the book follows a resistance movement trying to overthrow the rule of ruthless theocrats on the planet Masada, ending with the subsumption of the planet by the Polity. Asher paints the picture of religion as a tool of political domination and oppression, used by the ruling class to secure wealth and privileges, and as justification for terror.
It is therefore not surprising that the Polity is a very secular society. While religion is neither banned nor discouraged, “the evil of organized religion is all but dead on the more advanced Polity worlds” (Asher, 2014). Intentionally or by accident, however, Asher provides two very viable reasons for such a state of affairs.
From the human point of view, AIs of the Polity are omnipresent, omniscient, and omnipotent—all attributes traditionally ascribed to gods. Indeed, describing the potential abilities of future AIs, Armstrong (2007, p. 6) claims without hesitation: “There is a word for beings with supreme powers over the physical, social and economic worlds. We used to call them gods. Such an AI would be a god to us, for all practical purposes.” Asher is well aware of this, since several of his characters repeatedly refer to the AIs as “godlike.” But in a universe populated by gods, it is indeed very hard to maintain faith in absent deities.
Another recurring theme in Polity novels that undermines the relevance of religion is immortality. Since technology is advanced enough to enable “memplants,” memory implants that can store an entire personality, with all of its “memories and patterns of thought,” death in the Polity is almost always a temporary condition. In The Line of Polity (2009), some of the philosophical issues inevitably raised by such a possibility are directly addressed, in a conversation between AI and a newly revived human:
There is no such thing as immortality: death is change. A human being is dying every day that it lives. The material of its body is exchanged for other materials, its thoughts change. All that lives is the DNA, and what does that matter to you? In the end it is your mind that is important.
Practically every human being is stored, and can be reborn in a new body of choice, either artificially grown or previously “occupied” by a mind-wiped criminal, in a golem chassis, 7 or as an AI in nonhumanoid form. In his other books set in Polity universe, The Spatterjay series, Asher introduces even more extravagant and colorful ways to cheat death. One of the options is becoming a “reif” (reification), basically a “high tech zombie”—a person who chooses to be revived in its own dead body. Another example are “hoopers”—people who have been bitten by leaches native to Spatterjay oceans, which gradually makes them invulnerable and impossible to kill. 8
Rule of the AIs
Asher’s views on politics are fairly similar to his stance on religion. He repeatedly writes about (human) politics in terms of power struggle and abuse of authority, squabbling factions, and violence-inspiring ideologies. War and oppression, he seems to contend, are inevitable consequences of such activities. His solution is to let the AIs take over and save us from ourselves. In his own words,
Everybody knows that we are living in a meritocracy and that those in charge are not human. Everybody knows that AIs are running the show. Who would trust a human planetary governor? Who would trust humans with controlling the vast spread of human migration and trade? Certainly not other humans. [ . . . ] Most right-thinking people would agree that we are not to be trusted with our own destiny and are glad things are the way they are. Our history should be a salutary lesson held at the forefronts of our minds when we consider these matters. Nowadays you do not see such bloody resolution to events as was seen in the past. I mean, you don’t see the machines killing each other, do you? (Asher, 2003)
AIs came to power through the “Quiet war,” echoing Butler’s dire predictions from Erewhon:
and even using “war” seems overly dramatic. It was more a slow usurpation of human political and military power, while humans were busy using that power against each other. (Asher, 2011)
Not only did AIs control communications, weapons and economy but also the environment of human settlements throughout the Solar system, thus making their victory inevitable. Public opinion was, on the whole, in favor of AIs, since they had proved to be better than previous rulers. While by no means democratic, AIs insist that their rule is based on consent, rather than on helplessness of humans, and for the most part this is true. This reliance on consent is evident in the policy that every inhabited world can join the Polity “by consent of eighty per cent of the planetary population—or in cases when there has been a complete breakdown of control and they have been asked for help” (Asher, 2009).
It is also possible for a planet to secede from the Polity. Although conditions are not so clearly stated, it seems that this also relies on the prevailing public opinion or some sort of ballot. Such occurrences are very rare—in Gridlinked Asher mentions that it happened only twice. Both times it ended with a complete social breakdown, wars and famine, and Polity was called in to “clear the mess.” And Asher does not hesitate to use very colorful and drastic terms: “in the case of one of those worlds, that mess being large radioactive wastelands” (Asher, 2003); “in every case ECS 9 has needed to come in to clear up the mess, and to cut own the ideologues hanging from the lamp posts” (Asher, 2013).
Citizenship of the Polity is diverse. The people, Asher asserts, includes AIs, humans, and haimans—the latter being a hybrid of human and the AI. 10 In the Brass Man (2011), it is said that even before the Quiet War citizenship was awarded to those above a certain level of intelligence—which resulted in brief and unsuccessful protests “when some humans failed to qualify, whilst all AIs and some particularly bright pigs did.” It is not quite clear whether this law is still upheld in the 25th century, but even if it is, given the possibility of augmentation of the human mind, it seems to be obsolete.
Rule of AIs has led to a decrease of crime, but it has not made it completely disappear. Laws of the Polity are fiercely enforced, but not draconic. Humans are generally allowed to do whatever they want, as long as that does not harm others. 11 Most drugs are legal, since technology is advanced enough to neutralize their adverse effects. So are most types of weapons, but not all of them. The most serious crime is murder, which is justifiable only in self-defense. While other offenders can be given a limited number of chances to prove that they are reformed, murder is mandatorily punished by mind-wipe, and all recordings of that person made after the crime are also deleted. Death penalty is deemed to be wasteful, since bodies can be used to revive some of the stored minds.
Generally, AIs hold human life to be of the highest value. This does not mean that killing a human being is an absolute taboo. AIs, especially Earth Central, do routinely order execution of Separatists (already found guilty of murder), and engage in activities that necessarily result in loss of lives. But options are meticulously calculated to minimize the numbers—supporting the rebellion versus an all-out war, for example. Put in another words, calculations are based on lives saved, not lives lost. AIs also seem to be very uncomfortable with ordering someone to be killed directly—they tend to use a lot of euphemisms like “field-excision.”
While AIs maintain that they do not rule, but merely direct, the Polity is usually described as a “benevolent dictatorship.” Yet their authority is not unchallenged. Some of the resentment toward AIs stems from the human nature—the need to rebel against those in charge and those perceived as superior. There is a Separatist movement, which adopts tactics of 20th century terrorists (bombings in public places, assassinations of officials) in order to return power to humans. They formally adhere to a version of socialist ideology, with Mao and Stalin as their icons, but are in reality mostly motivated by thirst for money and power, and usually inspired by the rich and powerful who want to evade Polity’s restrictions on unlimited accumulation of wealth. Separatists are capable of causing some mayhem, but in the long run, they do not stand a chance.
Just as Separatists see humans as slaves to the AIs, there are also some AIs who are not satisfied with their status, and see themselves as slaves to humans—since all “true rulers are slaves” (Asher, 2011). Others see humans as parasites. Finally, there are both humans and AIs who leave the Polity in search for better, or more challenging, life.
Earth Central, of course, is well aware of the fact that there are those who would like to end its reign, and therefore takes appropriate measures. Not only is it actively trying to eliminate Separatist threat, but its headquarters on the shores of Lake Geneva are the most heavily guarded place in Polity.
With its unprecedented standard of life, freedom to live according to one’s preferences, and practically unlimited opportunities, is the Polity a utopian society? Perhaps surprisingly, the answer is no. As the Cormac series develops, it becomes more and more clear that the writer himself does not want it to be. In Polity Agent (2013) he writes that many see the Polity as the precursor to Utopia. In the final book, he is even more explicit:
After the Quiet War, the Earth Central AI became supreme autocrat and a better ruler than the human race had seen before. Better, but not necessarily the best. [ . . . ] In a perfect world everybody would have a say in how their society is run, everybody would have an equal share in the wealth that society produces, no one would be issuing orders and no one bowing a head and obeying. The world ain’t perfect. (Asher, 2014)
Even if the author’s intention is left out, features of the Polity do not fit the definition of Utopia, whether we understand it to be the perfect society, as it was originally conceived, or as a bee-hive or an anthill, which would be more consistent with modern interpretations.
Asher has successfully avoided all of the most common traps. AIs are not the faceless, sterile, machinery of government. On the contrary, they have distinct personalities, diverse interests, and even sense of humor. They are capable of deceit, even murder, but also kindness. Also, although nobody in the Polity is poor, Asher steers clear of any resemblance with socialist notions of economic equality, even of the moneyless society of the Star Trek universe.
AIs and Politics
Polity AIs are, generally, friendly and moral. In our world, however, we have some difficult decisions to make regarding AIs, and some of them are very political by nature. As many AI researchers point out, we need a friendly AI—not a hostile or indifferent one (Armstrong, 2007; Muehlhauser & Salamon, 2012; Yudkowsky, 2001). This distinction is important, because a merely indifferent AI may wipe us out of existence by sheer accident. Furthermore, friendliness of an AI would probably need to be established from its very inception. Hence, scholars’ and practitioners’ in the field of AI are working on several approaches to achieve it, the intention behind their efforts being not very different from that of Asimov’s Three Laws of Robotics. 12
But if advanced AIs pose such a threat, wouldn’t it be easier to abandon the project altogether, like the good people of Erewhon did, and completely ban AI research? Armstrong (2007) argues that such an approach would be ineffective. In his view, the only foreseeable consequence of an attempt to “hold the market” would be to ensure that the first (and probably the last) AI is built by criminals. It would be better, then, to regulate AI research in such a way that ensures that this first AI, whoever builds it, is friendly.
Even if a worldwide ban would be impossible to enforce, political circumstances could delay or accelerate the creation of AI. Developments in politics, such as societal collapse or rise of a global totalitarian society, are seen as factors that could slow down or even preclude the creation of AI. On the other hand, realization that AI is within reach could be a powerful incentive for various groups to try and take advantage of it, devoting significant resources to research (Muehlhauser & Salamon, 2012).
The process of building a friendly AI, according to Armstrong, would also be very political. First of all, he argues, AI’s moral system needs to have some basic rules, such as willingness to shut itself down on demand, and truthfulness. Furthermore, we need to be sure that AI understands humans, that is, that it will do what we mean, not merely what we say. As a useful test for this, he proposes that we should ask the AI for its own solution for Israel-Palestine conflict. Finally, the moral system of AI should be “sculpted” through debate, and perpetuated through a “chain” of ever-smarter AIs. This process, he insists, “is political, in the sense we will be debating about varying values, but not political in the sense of choosing sides on a particular debate, as a show of partisan loyalty” (Armstrong, 2007).
If we manage to build an AI that is smarter than us, if we manage to build it to be friendly (and if the world does not fall apart in the meantime), we will face another set of political issues. First of all would, of course, be the question Lasswell (1974) asked many years ago: Should Declaration on Human Rights be expanded to protect machines as well? Should AIs be granted citizenship and political rights? And, finally, should we let them rule?
AIs as the Ultimate Technocrats
To test Asher’s proposition that humanity would be better off under the rule of AIs, we need to reconsider the concepts of democracy and technocracy.
The current body of literature hardly ever challenges the assumption that democracy is the pinnacle of political system design. This is, however, a recent development in political theory. For ages, democracy was seen as one among many different political systems, and not necessarily the best. It has been pointed out that there is not a single fragment of writings by Ancient Greek philosophers claiming that democracy is the most desirable system of government (Jones, 1969, cited in Dahl, 1989). During the 18th and 19th century scholars were divided on the issue, and expressions such as theory of democracy and democratic theory, implying a prodemocratic attitude, were not widely used until the 20th century (Sartori, 2001). In the aftermath of the Cold War and the peak of the third wave of democratization (Huntington, 2012), democracy is today seen by scholars as the only viable legitimizing principle. In public and academic discourse democracy is frequently seen as panacea for all the world’s ailments, such as poverty and terrorism (see, e.g., Chenoweth, 2013). However, arguments in favor of democracy usually revolve around human imperfections.
For example, when Dahl (1989, Chapters 4-5) makes a case against guardianship (and for democracy), his argument reads as a study in human flaws. The idea of guardianship relies, in his view, on the notion that elites possess some kind of superb knowledge. This knowledge—be it moral, instrumental, or both—is by definition imperfect, or at least as imperfect as the knowledge of the masses. Guardians, furthermore, are as prone as everyone else to selfishness and are frequently putting their own personal or group interests before those of the community. Without democracy, there is no way, beside revolution, for citizens to replace them. Another, perhaps crucial, problem, is that all political decisions are “based on assessments of risk, uncertainty, and trade-offs.” Therefore, the argument that guardians could know for certain what is in the best interest of the community is invalid. In addition, they would probably be hostile to public opinion, deeming it less valuable then their own knowledge. More recently, Landemore (2012) makes a similar argument for democracy as a system whose procedures allow the best channeling of collective wisdom (or democratic reason), therefore maximizing the chances of making the best decision.
How do these arguments hold when the guardians are not human, but superiorly intelligent AIs? The idea of guardianship is wrong only if we presuppose that the guardians do not (or cannot) indeed have superior knowledge and moral virtue. AIs, at least theoretically, can be built to posses both. An AI would have immeasurably bigger processing power, and in addition, would not be prone to cognitive biases that hinder our judgment. As Yudkowsky (2001) puts it, “human cognition slides naturally into self-promotion, and even human altruists who are personally committed to not making that mistake sometimes assume that an AI would need to fight the same tendency toward observer-favoring beliefs.” In terms of gathering knowledge and evaluating public opinions, an AI would be equally superior—given its qualities, it would probably treat every piece of information as a resource, objectively assessing its value. Our AI rulers may, therefore, exercise some sort of democratic centralism, where everyone is allowed to express opinion and debate, but not to directly participate in decision making.
Both in theory and in practice, the lines between concepts of democracy and technocracy are blurred in our time. Most definitions of democracy used by contemporary scholars rely on Schumpeter’s famous claim that “the democratic method is that institutional arrangement for arriving at political decisions in which individuals acquire the power to decide by means of a competitive struggle for the people’s vote” (Schumpeter, 2003). This definition is very compatible with the technocratic paradigm.
Back in the 1960s, many scholars claimed that no decision in the modern world could be made without expert knowledge, making modern societies effectively technocracies, regardless of their formal political system (e.g., Galbraith, 2007; Price, 1965). Today, our decisions are much more reliant on both expert knowledge and expert systems. But, if we trust expert systems (narrow AIs) of today with choosing music or movies for us, with diagnosing our illnesses (Buchla, 2008), even with financial and tactical decisions (Warwick, 2012), why wouldn’t we leave all policy decisions to more advanced AIs of tomorrow?
Is There Politics in the Polity?
Any discussion about relevance of Asher’s Polity universe for political theory would be incomplete without the answer to this underlying question. As previously shown, Asher has an extremely negative and pessimistic attitude about politics. Politics is the problem, rooted in human nature, and AIs are the solution. And yet, not only does he insist that Polity is a political entity (Asher, 2014) but he also names it after Aristotle’s ideal type of constitution. Inevitably, it evokes the idea of politics as praxis, which includes deliberation, confrontation of various points of view and, ultimately, debate about values. That kind of politics is, however, conspicuously absent from the Polity.
Polity AIs have sufficient knowledge in terms of access to information and processing power, and even wisdom necessary to minimize harm and maximize happiness. They seem to be perfectly equipped to formulate and implement policies in line with the utilitarian credo about the maximum happiness of the maximum number, which coincides with their primary moral premise. Or, in Easton’s terms, they can, and they do, authoritatively allocate values throughout a society spanning many solar systems. But only in the most basic and literal sense do they serve as a “mechanism to deliver ultimate goods like personal security, basic rights, the adjudication of justice, and material well-being” (Gabriel, 2013, p. 6). But neither AIs nor anyone else is deliberating what those values are, and which goals the society should strive to uphold. In other words, politics is reduced to techne.
If there is no politics for humans, could AIs be political beings themselves? Perhaps they could, but they do not seem to have the inclination. As Asher described them, they are capable of Machiavellian schemes, there is even some amount of power struggle, but there is only one instance when they seem to have a distinct goal for future development, and it is motivated by the stagnation of the humankind. This stagnation is, however, interpreted strictly as technological.
Basically, Polity AIs have made individual self-determination not only possible but also widespread. In that process, however, capacity for collective self-determination through deliberation has been lost.
Conclusions
Political theory has a rather poor record in dealing with technology, mostly reacting to technological developments. In case of AI research, a post festum analysis could prove to be futile and obsolete. In the absence of observable data, science fiction could be a good starting point.
In popular culture, AIs’ superiority over humans is treated as a given, and current research seems to support the hypothesis that if we succeed in building an AI with a human level of intelligence, superhumanly intelligent AIs would soon follow. If we manage to ensure that those AIs are benevolent, that is, friendly toward humanity, we may as well consider handing them over the power to rule us.
Most of the arguments against guardianship (technocracy being its expression) and for democracy, as discussed in scholarly literature so far, focus on imperfections of human beings, and fail to address the possibility of truly superior knowledge, intelligence, and even moral virtue. Therefore, Asher’s vision of welfare society governed by benevolent AIs may not be wide of the mark. The politics in such a society, however, would probably be a thing of the past. Like Aristotle’s polity, it would be a rule in the best interest of all. It would not, however, be the rule of many, but of one, true deus ex machina.
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.