The historical evolution of legal thought and artificial intelligence
04 july 2026
By Kyriakos Kyriazopoulos
"The historical evolution of legal thought and artificial intelligence"
Introductory remarks
The emergence of artificial intelligence undoubtedly constitutes one of the most significant developments of the contemporary era. The speed with which algorithmic systems are entering public administration, the administration of justice, the economy, science, and decision-making more generally has created the impression that humanity is confronted with an entirely new phenomenon, one that requires the development of wholly new legal instruments and unprecedented theoretical approaches. Indeed, the discussions surrounding the liability of autonomous systems, the protection of personal data, algorithmic transparency, the prohibition of discrimination, the automation of administrative functions, and the support of judicial decision-making through artificial intelligence systems appear, at first sight, to constitute an entirely new field of knowledge, detached from the historical evolution of legal science.
A more careful historical and philosophical examination, however, leads to a different conclusion. Artificial intelligence does not constitute merely another technological innovation affecting the application of the law. Rather, it functions as the occasion for the re-emergence of fundamental questions that have accompanied legal thought ever since its inception. Behind every discussion concerning algorithms, large language models, machine learning systems, or automated decision-making procedures, the classical problems of the philosophy of law once again emerge: what is the nature of law, from where does it derive its binding force, what is its relationship with political authority, what is the role of human judgment, what is the significance of interpretation, how is legality distinguished from justice, and to what extent technical rationality may replace human practical wisdom.
It is precisely for this reason that artificial intelligence should not be approached exclusively as a new subject of positive law, nor merely as a field of technological regulation. Its most substantial significance lies in the fact that it brings back to the centre of theoretical discussion the most fundamental questions in the history of legal thought. Viewed in this light, historical inquiry does not constitute merely an introductory retrospective or a didactic reference to the great milestones of legal science. On the contrary, it constitutes the indispensable interpretative horizon within which an understanding of today's challenges becomes possible.
The present monograph is founded precisely upon this fundamental premise. Rather than examining artificial intelligence as an isolated subject, it seeks to reconstruct the diachronic course of legal thought from the earliest forms of social normativity to the most contemporary theories concerning human rights and algorithmic governance. This historical reconstruction does not aim merely at presenting the major schools of legal philosophy or descriptively enumerating theoretical currents. Its objective is to demonstrate that every significant period in the history of law formulated answers to problems that today re-emerge in a different form as a consequence of artificial intelligence.
The evolution of legal thought reveals that the history of law is not a linear accumulation of knowledge but a continuous dialectic between stability and change. Every era reformulates the same fundamental questions by employing a different historical vocabulary and different institutional instruments. Thus, whereas archaic societies sought ways of limiting violence through normative forms of social cohesion, classical Greek thought investigated the relationship between nature and law, Roman jurisprudence developed the technical structure of legal reasoning, the Christian and medieval tradition reintroduced the concept of higher justice, modernity highlighted sovereignty and individual rights, the Enlightenment advanced the rational universality of law, positivism sought to establish the validity of legal rules, legal realism directed attention to judicial practice, while contemporary interpretative and critical theories emphasized the importance of principles, legal reasoning, and social relations of power.
Artificial intelligence enters precisely into this centuries-long historical discussion. It does not create a new universe of concepts but reshapes the environment within which concepts already familiar to the legal tradition are called upon to operate. The relationship between the legal rule and the concrete case, the distinction between formal and substantive justice, the significance of legal reasoning, the concept of responsibility, the public accountability of authority, the protection of human dignity, and the necessity of institutional oversight acquire a new practical dimension, without ceasing to constitute the very same fundamental theoretical categories that have shaped the history of legal science.
The present study therefore adopts a methodological approach that differs from a considerable part of the contemporary literature. It does not seek merely to identify historical "precursors" of artificial intelligence, nor to present a history of technology. On the contrary, it employs the diachronic evolution of legal thought as an interpretative instrument for understanding contemporary problems. Historical knowledge thus acquires a theoretical function. It makes it possible to recognize that contemporary controversies concerning algorithmic decision-making, the use of predictive systems, the automation of public administration, or the application of large language models in legal practice do not constitute isolated technological issues, but rather specific manifestations of deeper controversies running throughout the entire history of legal theory.
Viewed from this perspective, artificial intelligence functions as a catalyst revealing the internal tensions of legal thought itself. It once again brings to the forefront the opposition between formalism and practical wisdom, between positivism and natural law, between systematic standardization and interpretative creativity, between functional efficiency and substantive justice, and between technical rationality and human responsibility. In this way, it becomes evident that the real challenge does not consist merely in adapting the law to new technologies, but in renewing the theoretical self-consciousness of legal science itself.
The history of legal thought demonstrates that every major technological, social, or political transformation has compelled the law to reconsider the very conditions of its own existence. The same is true today. Artificial intelligence does not negate the need for legal theory; on the contrary, it renders necessary a deeper return to its historical and philosophical foundations. Only if the long evolutionary course of the concept of law is properly understood can it be correctly assessed to what extent algorithmic technology serves as an instrument for the administration of justice, and at what point it begins to threaten the very fundamental conditions of the legal order.
With this methodological point of departure, the present monograph begins. It will trace, in a diachronic manner, the evolution of legal thought from the earliest forms of normativity to the most contemporary issues of artificial intelligence, seeking to demonstrate that the history of law does not constitute the past of legal science but its indispensable theoretical future.
CHAPTER I
The Earliest Forms of Normativity and Law as a Social Technology of Order
The history of legal thought does not begin with the formulation of abstract theories concerning law, nor with the systematic philosophical investigation of the concept of justice. Long before the emergence of philosophy, legislative bodies, and organized state mechanisms, human communities had already developed forms of normative organization which secured the cohesion of collective life and limited the diffuse violence that characterizes every society in which stable institutions have not yet been formed. These first forms of normativity did not yet constitute law in the modern sense of the term. There was no clear distinction between legal rule, moral imperative, religious command, customary practice, and political authority. All these dimensions constituted a single normative reality, within which the preservation of the community was the supreme purpose.
Archaic society did not understand transgression as the mere violation of a rule of conduct. Transgression had a much broader significance. It disturbed the equilibrium of the entire community, created a state of impurity, caused a rupture with the tradition of the ancestors and, at the same time, was considered to disrupt the very relationship of society with the sacred. For this reason, the restoration of order was not achieved only through the imposition of sanctions but also through rituals of expiation, procedures of reconciliation, or forms of restoration of collective equilibrium. Law is therefore born not as an abstract system of logical rules but as an institution for the management of social conflict.
This observation has particular significance for the contemporary theory of artificial intelligence. The impression is often presented that algorithms constitute merely advanced tools of calculation, which improve the effectiveness of administrative or judicial procedures. Historical experience, however, shows that every technology for organizing social life is at the same time also a technology for producing normativity. It is not confined to the application of pre-existing rules but actively contributes to the determination of what is considered normal, permissible, dangerous, or deviant. In this sense, artificial intelligence is not only a means of calculation but also a new mechanism for the constitution of social order.
The history of early societies makes clear that every form of normative organization always contains a specific conception of the human being and of the community. Archaic societies did not merely seek the prevention of violence. They sought the preservation of a specific form of social cohesion, which rested upon common tradition, kinship, collective memory, and shared religious experience. Normativity had not yet acquired autonomy; it constituted an expression of the entire social and cultural organism.
An analogous function is observed, under different historical conditions of course, also in contemporary algorithmic systems. Every machine learning model, every system of data classification, and every mechanism of risk assessment rests upon prior choices concerning which data are considered significant, which variables will be used, which categories will be created, and which result will be characterized as desirable. These choices do not constitute technically neutral acts. On the contrary, they incorporate evaluative presuppositions, social hierarchizations, and political priorities. Just as archaic societies defined the boundaries between order and disorder through their customs and rituals, so too contemporary algorithmic systems contribute to the determination of the boundaries between the normal and the deviant, between the reliable and the dangerous, between the permissible and the unacceptable.
This analogy certainly does not mean that contemporary algorithms constitute a simple repetition of archaic institutions. It means, however, that both archaic forms of normativity and contemporary digital technologies perform a common function: they organize social reality through categorizations, classifications, and procedures of control. Technology changes; the fundamental normative function remains impressively stable.
This development acquires even greater significance when written legislation appears. The transition from an exclusively oral tradition to the public recording of rules constitutes one of the most significant ruptures in the entire history of law. Writing allows the rule to acquire stability, duration, and publicity. The law ceases to depend exclusively upon the memory of the elders, priests, or leaders of the community and is transformed into a public point of reference, accessible to all those who belong to the political community.
The significance of this transformation is not exhausted in the greater legal certainty offered by the written form. Writing makes possible systematic analysis, comparison, interpretation and, ultimately, the development of legal science. Without the stability of the written text, neither systematic doctrinal elaboration nor the development of conceptual categories could exist. The history of legal science is inconceivable without the prior history of legal writing.
This same historical precondition today constitutes the basis for the functioning of artificial intelligence systems. Contemporary systems of natural language processing, mechanisms of legal data mining, legal analytics applications, and large language models presuppose the existence of a vast body of texts, legislation, case law, and theory, which can be subjected to processing. In other words, artificial intelligence does not create its object out of nothing. It is based upon a historical process of many millennia, during which law was gradually transformed into a systematized body of texts, concepts, and normative structures.
This historical observation, however, also leads to a second, equally significant consequence. The formalization of law has never been identified with the realization of justice. The existence of written laws has never excluded social inequality, political arbitrariness, or institutional oppression. On the contrary, many times the very systematicity of rules was used for the legitimation of profoundly unequal social relations. History therefore teaches that the clarity, stability, and consistency of rules constitute necessary but not sufficient conditions of justice.
Exactly the same is true of artificial intelligence. The capacity to produce consistent results, the uniform application of predetermined criteria, and the statistical accuracy of predictions are not sufficient by themselves to establish the normative correctness of a decision. An algorithmic system may be entirely consistent and at the same time profoundly unjust, if the presuppositions upon which it was constructed reflect historical prejudices, social inequalities, or erroneous evaluative assumptions. The history of the first forms of normativity thus reminds us of a fundamental truth, which runs through the entire evolution of legal thought: every technology for organizing social life, however advanced it may be, must be evaluated not only on the basis of its effectiveness but primarily on the basis of the kind of justice it serves.
Thus, already from the beginnings of the history of law, a pattern emerges that will accompany all subsequent theoretical developments. Law is never a mere technique for managing social reality. It is always at the same time a mode of constituting the community, a form of organizing authority, and an expression of a specific conception of the human being. This observation will acquire even deeper philosophical content when Greek thought transforms law from practical social experience into an object of systematic theoretical reflection, inaugurating a new era in the history of legal philosophy.
CHAPTER II
The Greek Turning Point: Physis, Nomos, and the Birth of Legal Philosophy
The transition from the archaic forms of normativity to Greek philosophical thought constitutes one of the deepest turning points in the history of law. For the first time, law ceases to be treated exclusively as an element of tradition or as an inseparable part of religious and political life and is transformed into an object of autonomous theoretical reflection. Greek philosophy is no longer content with the description of the rules in force or with the reproduction of established institutions. It raises the question why rules are binding, what their relationship to nature is, what their origin is, whether they express true justice or merely the power of dominant groups, and, ultimately, what the limits of human legislative authority are. In this way, legal philosophy is born as an autonomous form of theoretical inquiry.
The significance of the Greek turning point is not confined to the history of ideas. The questions formulated during the classical period continue to constitute the core of contemporary discussions on artificial intelligence. The controversy as to whether algorithms are neutral or whether they incorporate pre-existing evaluative choices, the discussion concerning the relationship between technical effectiveness and substantive justice, as well as the problematique concerning the possibility of automated decision-making are, in reality, new forms of old philosophical controversies which had already appeared during the period of classical Greek thought.
The first great turning point is brought about by the Sophists. The Sophistic movement moves away from the conception according to which laws constitute the self-evident expression of a transcendent order and highlights their historical and conventional character. The distinction between physis and nomos makes it possible, for the first time, to pose in a systematic manner the question whether the rule in force truly corresponds to justice or whether it is merely the product of political decision. The relativity of laws, their differentiation among different cities, and their dependence upon specific social and political conditions lead to the conclusion that normativity can no longer be regarded as self-evident or undisputed.
This Sophistic approach acquires exceptional topicality in the age of artificial intelligence. Algorithms are often presented as neutral mathematical tools, free from the weaknesses and prejudices of the human being. The philosophical experience of the Sophists, however, reminds us that no form of normative organization is free from the conditions within which it is produced. Just as human laws constitute the product of specific social choices, so too algorithmic models incorporate decisions concerning the data that will be used, the categories that will be created, the variables that will be evaluated, and the results that will be considered desirable. The mathematical form of an algorithm does not negate the historical character of its presuppositions. On the contrary, it often makes their recognition more difficult.
Sophistic thought, therefore, introduces an element of critical self-consciousness, which remains entirely topical. Artificial intelligence should not be regarded as a natural phenomenon or as an objective mechanism for the production of decisions. It constitutes a human creation, incorporates human choices, and functions within specific social and institutional structures. The recognition of this reality constitutes a necessary precondition for every serious theory of algorithmic justice.
In opposition to the relativistic tendency of the Sophists, Socrates shifts the discussion to a different level. His interest is not focused primarily on the historical origin of laws but on their relationship with the conscience and moral responsibility of the human being. His stance toward his own condemnation constitutes one of the most characteristic moments in the entire history of political and legal philosophy. His choice not to escape from prison, despite the possibility offered to him, expresses the conviction that the legal order is not merely an external mechanism of coercion but a relationship of mutual commitment between the citizen and the community.
The Socratic problematique acquires new significance when transferred to the field of artificial intelligence. In contemporary societies, the citizen is called upon ever more frequently to comply with decisions that are produced or supported by complex algorithmic systems. The question therefore arises as to what the object of obedience is. Does one obey human judgment, state authority, an institutionally established procedure, or a mathematical prediction produced by a machine learning system? The Socratic tradition reminds us that legality is not exhausted by the effectiveness of the procedure. It always presupposes a form of moral and political legitimation, without which obedience is transformed into mere submission.
The theoretical culmination of Greek thought takes place in Plato. For Plato, justice is never identified with the simple application of predetermined rules. It constitutes a state of harmony both of the soul and of the polity. Law possesses a pedagogical mission; it aims at the formation of the virtuous citizen and not only at the regulation of external behaviours. This conception introduces a critical dimension that is often absent from contemporary technocratic discourse concerning artificial intelligence.
The evaluation of an algorithmic system cannot be exhausted in the examination of its accuracy or its speed. It must also extend to the question of what purpose it serves and what form of political and social order it contributes to shaping. A system that is extremely effective in predicting criminal behaviour or in allocating administrative resources cannot automatically be considered just if the aims it serves conflict with human dignity, equality, or democratic autonomy. Platonic theory indicates that technical correctness is not sufficient to establish normative correctness.
The deeper contribution of Greek philosophy to the contemporary theory of artificial intelligence, however, comes from Aristotle. Aristotelian theory of law gathers almost all the concepts that continue to constitute the core of today’s discussion. The famous characterization of law as “reason without passion” initially appears to support the idea of an impersonal, logical, and consistent application of rules. Indeed, the proponents of the extensive use of artificial intelligence often advance the argument that algorithmic systems can limit human bias, eliminate emotional influences, and ensure greater uniformity in decision-making.
The very same Aristotelian theory, however, simultaneously introduces perhaps the strongest argument against the complete automation of legal judgment. The concept of equity constitutes the mechanism through which Aristotle recognizes that every general rule inevitably presents limits. No general formulation can foresee all the particularities of factual circumstances. Because of this incapacity, practical wisdom is required, namely the capacity of the judge to ascertain when the strict application of the general rule leads to a result contrary to justice.
Equity does not constitute an exception to the law nor an expression of arbitrary leniency. On the contrary, it constitutes a higher form of application of the law itself, because it restores justice where the unavoidable generality of the rule creates injustice. This distinction acquires exceptional significance in the age of artificial intelligence. Contemporary computational systems can incorporate ever more variables, recognize complex patterns, and continuously improve their predictive capacity. However, however advanced they may become, they continue to function through categorizations and generalizations. Equity, by contrast, presupposes the capacity to recognize the uniqueness of the concrete case and to evaluate its significance within the overall normative framework.
It is precisely here that one of the deepest philosophical limits of algorithmic justice is revealed. Practical wisdom does not consist in the simple processing of more information nor in the increase of computational power. It constitutes a form of judgment, which combines experience, evaluative assessment, institutional understanding, and responsibility toward the concrete human person. Greek philosophy thus arrives at a conclusion that continues to constitute perhaps the most significant theoretical limit of every attempt at the complete automation of law: the logic of the rule is necessary for the functioning of the legal order, but justice is completed only when the generality of the rule encounters the human judgment that can discern the uniqueness of the concrete case.
With this legacy, the Greek foundation of legal philosophy is completed. The concepts of physis, nomos, justice, equity, practical wisdom, and the relationship between rule and concrete case will henceforth constitute the stable theoretical background of every subsequent development. The next great historical period will transfer these philosophical achievements into the field of systematic legal science, where the Roman tradition will transform law into a complete technique of legal reasoning and conceptual organization.
CHAPTER III
Rome and the Birth of Legal Science: Law as a Technique of Reason and Interpretation
Greek philosophy founded the great concepts of justice, nature, law, and the political community. The Roman tradition, by contrast, was not primarily interested in formulating a comprehensive philosophy of law. Its distinctiveness lies in the fact that it transformed law into an autonomous science, with its own particular methodology, strict conceptual organization, and exceptionally developed techniques of interpretation. Whereas the Greeks primarily investigated the question of what law is, the Romans developed the question of how law is applied in a systematic, consistent, and logically controllable manner. This transition constitutes one of the most significant moments in the history of legal thought, because it creates the preconditions for the development of legal science as a distinct field of knowledge.
The Roman legal tradition is characterized by a remarkable balance between practical necessity and theoretical abstraction. The great jurists do not treat law either as a simple political will or as a pure philosophical idea. They understand it as ars, as art, that is, as a form of scientific knowledge that requires methodology, discipline, logical consistency, and the continuous exercise of judgment. The famous concept of ars iuris expresses precisely this synthesis. Law is not automatism nor accidental practice. It is the technique of correct judgment.
This choice proved decisive for the entire subsequent European legal tradition. The creation of abstract concepts, the development of general categories, the distinction between different types of rights, obligations, and legal relationships, the elaboration of systematic methods of interpretation, and the organization of legal argumentation constitute a conceptual edifice which continues to this day to form the foundation of most continental legal orders.
The significance of this development for contemporary artificial intelligence is exceptionally great. The functioning of every advanced system of legal data processing presupposes that its object possesses sufficient conceptual organization. An algorithm cannot process an entirely amorphous set of social information. It presupposes the existence of categories, definitions, conceptual relations, and relatively stable structures. In other words, artificial intelligence functions effectively precisely because the historical evolution of law, and especially Roman legal science, gradually transformed law into a systematized field of knowledge.
This observation often leads to the impression that Roman legal science constitutes the direct precursor of contemporary algorithmic applications. Indeed, there are evident analogies. The Roman method rests upon classification, conceptual analysis, logical consistency, and the systematic elaboration of legal relationships. The same characteristics are pursued by contemporary legal analytics systems, legal expert systems, and many artificial intelligence applications used in legal research.
This analogy, however, must not lead to misunderstanding. Roman ars iuris is never identified with mechanical formalism. On the contrary, the technical dimension of law always remains inseparably connected with human interpretation. Roman jurists do not consider that rules are applied automatically. They understand that every concrete case requires evaluation of the facts, selection of the appropriate category, comparison with previous cases, and ultimately the formation of a syllogism that can be logically justified.
The significance of interpretatio is characteristic. Interpretation is not considered a secondary function, but an inseparable part of legal science itself. Law is not exhausted in the written text. Nor does its application consist in the simple matching of rules and facts. The jurist is called upon to understand the purpose of the rule, to evaluate its relationship with other provisions, and to integrate the concrete case into the overall logic of the legal order.
Precisely here an essential difference is located between the Roman legal tradition and certain contemporary forms of algorithmic processing. Most computational models tend to treat legal information as a set of data to be classified. This classification may be exceptionally complex and may rest upon millions of examples. It remains, however, essentially a process of statistical pattern recognition. Roman interpretatio, by contrast, is always an act of normative argumentation. It does not merely aim at the prediction of a result but at its justification.
This distinction acquires particular significance when examining the reasoning of judicial decisions. In Roman thought, the correctness of a solution does not depend only on the final conclusion but also on the logical path that leads to it. The ratio decidendi constitutes the core of legal reasoning. It expresses the reasoned connection between rule and concrete case and permits the public control of the correctness of the decision.
This requirement remains fundamental also in the contemporary era. A system of artificial intelligence may arrive at a result which statistically appears highly probable. The question, however, posed by legal science is not only whether the result is probable or effective. It is whether it can be reasoned in an institutionally controllable manner. Reasoning constitutes an inseparable element of legality. It is not a subsequent addition to the decision but an essential part of the judicial function itself.
The Roman experience thus allows us to understand better yet another contemporary problem, which is often characterized as the problem of the “explainability” of algorithmic systems. In public discourse, explainability is often presented as a technical characteristic of algorithms. In reality, however, it constitutes above all a legal requirement. The legal order is not satisfied with the correct functioning of a mechanism. It requires that this functioning be capable of being reasoned in a manner that permits challenge, re-examination, and judicial control.
From this perspective, the history of Roman legal science acquires exceptional topicality. Its great contribution does not lie only in the creation of a systematic body of concepts but chiefly in the establishment of the idea that legal technique is always a technique of public reasoning. The jurist is not limited to finding solutions; he or she must explain why the specific solution is justified within the system of law.
This difference allows a more general observation to be formulated regarding the relationship between technique and justice. The technical dimension of law is undoubtedly necessary. Without conceptual precision, without logical consistency, and without methodological discipline, the legal order would end in an arbitrary exercise of authority. Technique, however, does not constitute an end in itself. It always serves a broader normative pursuit, namely the realization of justice through institutionally controllable procedures.
The legacy of Roman legal science thus acquires particular significance for every contemporary theory of artificial intelligence. It reminds us that the systematization of law constitutes a necessary precondition of algorithmic processing, but at the same time it reveals the limits of every technological approach that attempts to replace interpretation with automatism. Law is certainly technique; it is, however, the technique of human judgment and not a mechanism of mechanical production of results.
This historical synthesis will undergo a new transformation with the appearance of Christianity and medieval thought. There, the concept of technical correctness will once again be subordinated to the search for a higher measure of justice, through the theory of natural law, the common good, and the moral foundation of political authority.
CHAPTER IV
Christian and Medieval Thought: Natural Law, the Common Good, and a Higher Measure of Justice
The transition from Roman legal science to Christian and medieval thought does not constitute a simple historical succession nor the replacement of one system of ideas by another. On the contrary, it constitutes a creative synthesis, within which the high technique of the Roman legal tradition encounters the deep ethical and theological problematique of Christianity. This encounter substantially changes the way in which law is henceforth conceived. The concept of legality ceases to be exhausted in institutional validity or in the systematic organization of rules and is integrated into a broader conception of justice, the human person, and the common good. This transformation will exert a profound influence upon the entire Western legal tradition and continues to constitute, even today, one of the most important theoretical axes of the discussion concerning artificial intelligence.
Christian thought introduces into law a dimension that was absent from most earlier conceptions. The person is no longer treated merely as a member of the political community nor exclusively as a bearer of rights and obligations. He is recognized as a being endowed with intrinsic value, freedom, and conscience. The relationship between law and the human being thus acquires a personal character. Obedience to the law does not constitute only conformity to external rules but is connected with the inner responsibility of conscience.
This new perspective becomes apparent already in the work of Augustine. The famous position that a state without justice does not differ essentially from an organized band of robbers is not merely a theological declaration. It expresses a deep theoretical shift. Political power is not automatically legitimized by its effectiveness nor by its capacity to impose obedience. Legality depends upon the relationship of authority to justice. Therefore, law cannot be identified with any expression whatsoever of organized state will. There is always a higher criterion, on the basis of which the legal order itself is subject to evaluation.
This position presents remarkable topicality in the age of artificial intelligence. Contemporary algorithmic infrastructures often appear as neutral means of optimizing public administration, justice, or economic activity. Yet Augustinian thought reminds us that effectiveness is not sufficient to legitimize a form of authority. A system may function with exceptional accuracy, process vast volumes of data, and produce consistent decisions; if, however, these decisions violate human dignity or perpetuate forms of injustice, then technical excellence is not sufficient to establish their legitimacy.
This theoretical direction acquires full systematic formulation in the work of Thomas Aquinas. Aquinas attempts to unify Aristotelian philosophy with Christian theology and creates one of the most influential theories of law in European history. The distinction between eternal law, natural law, human law, and divine law does not constitute a simple classification of different forms of normativity. It expresses an entire conception of the relationship between enacted rules and objective justice.
Human law acquires validity not only because it was enacted by the competent political authority but because it aims at the common good and agrees with the dictates of right reason. If it departs from these presuppositions, then it may perhaps continue to retain the form of law, but it ceases to correspond to its essence. Aquinas’s famous formulation that an unjust law constitutes a distortion of law does not mean a denial of institutional order; it means that the concept of law always includes an evaluative dimension, which cannot be eliminated.
The significance of this medieval tradition for the theory of artificial intelligence is particularly great. In contemporary normative discourse, concepts constantly appear such as human oversight, transparency, explainability, non-discrimination, the protection of human dignity, the possibility of challenging decisions, and the accountability of artificial intelligence systems. Often these concepts are presented as exclusively contemporary requirements arising from rapid technological development. In reality, however, they constitute more recent expressions of a much older theoretical tradition, according to which the validity of a form of authority does not depend exclusively on the effectiveness or legality of the procedure but also on its agreement with higher principles of justice.
This observation acquires particular significance when the issue of the autonomy of algorithmic systems is examined. The more decision-making procedures are transferred to complex machine learning systems, the more intensely the temptation appears to evaluate correctness exclusively on the basis of functional success. If the system predicts effectively, if it reduces errors, if it increases the speed of procedures, then it is considered successful. Medieval theory, however, insists that every form of authority must also be evaluated in light of the purpose it serves. It is not sufficient to examine the manner of its functioning; it is also necessary to examine the direction toward which it leads the political community.
The concept of the common good acquires central significance here. In contemporary technocratic thought, the conception often prevails that the optimization of procedures constitutes, in itself, a sufficient goal of public administration. Yet the common good is not identified with the greatest possible effectiveness. It concerns the overall quality of common life, the preservation of freedom, the protection of the dignity of the person, equality before the law, social cohesion, and trust in institutions. An algorithmic system may improve certain quantitative indicators of the functioning of administration and at the same time undermine deeper presuppositions of the democratic polity.
From this point of view, medieval thought functions as a counterweight against contemporary technocracy. It does not deny the value of technique nor does it underestimate the significance of the rational organization of institutions. It reminds us, however, that every technique must serve a broader normative purpose. Technology cannot become autonomous from the ethical and political evaluation of its consequences.
The same problematique also appears in the issue of human oversight. The requirement that human intervention in decision-making procedures must always remain possible does not constitute a simple guarantee of safety against technical errors. It expresses a deeper anthropological presupposition, according to which the final responsibility for the justice of decisions cannot be transferred completely to impersonal mechanisms. The legal order always presupposes the existence of responsible persons who assume accountability for their decisions.
The Christian and medieval tradition, therefore, adds a new fundamental dimension to the history of legal thought. Legality is not exhausted either in the existence of rules or in the logical coherence of the legal order. Neither technical excellence nor systematic organization constitutes, in itself, a sufficient criterion of justice. Above formal validity there always exists the question of purpose, of the dignity of the human being, and of the common good.
This theoretical legacy will be transformed once again during modernity. The emergence of the modern state, the theory of sovereignty, the development of the social contract, and the gradual foundation of individual rights will radically alter the framework within which the legal order is henceforth conceived. The questions, however, formulated by medieval thought will continue to accompany every subsequent theory of law, reminding us that no technical form of organizing society can be considered self-sufficient when it is cut off from the justice it is obliged to serve.
CHAPTER V
Modernity: Sovereignty, Rights, and the New Foundation of the Legal Order
The transition from the medieval to the modern era marks one of the deepest transformations in the history of legal thought. The dissolution of the unity of the medieval world, the formation of the sovereign state, the religious wars, the emergence of the scientific method, and the gradual formation of the first liberal theories radically change not only the structure of political authority but also the way in which the very concept of law is understood. Whereas the medieval tradition placed at the centre the common good and the subjection of human law to a higher normative order, modernity seeks a new foundation of the legal order through the concept of sovereignty, individual freedom, and the social contract.
This transformation does not constitute the simple replacement of one theoretical scheme by another. It represents a change in the entire way of understanding the political community. Medieval society understood law as part of an objective moral order, which pre-existed state authority. Modern thought, by contrast, seeks to explain how the constitution of a legal order is possible in a world where religious and moral unity has now been ruptured. The question shifts from the search for objective justice to the search for stable institutions capable of ensuring peace, security, and predictability.
Hobbes’s theory constitutes the first great systematic answer to this new problem. The experience of civil war leads the English philosopher to the conclusion that without strong political authority human society inevitably leads to a state of mutual insecurity. Law thus acquires a new mission. It is no longer primarily an expression of virtue nor a mechanism for the perfection of the human being. It is transformed into an institutional means of limiting conflict and guaranteeing collective survival.
This Hobbesian view presents striking analogies with many contemporary applications of artificial intelligence. A large part of the algorithmic systems used today in public administration, policing, risk management, or the prediction of criminal behaviour rests precisely upon the pursuit of security through predictive analysis. Their logic is characteristically preventive. Instead of merely addressing already manifested violations, they attempt to predict future behaviours and to reduce uncertainty through statistical assessments.
This historical correspondence is not accidental. Artificial intelligence significantly reinforces a conception of law according to which the primary purpose of the legal order is the effective management of social risk. The more accurate prediction becomes, the stronger appears the temptation to shift the centre of gravity from the administration of justice to the management of the probability of future violations. Law thus risks being transformed from a system for judging concrete acts into a mechanism of continuous assessment of possible behaviours.
This development highlights one of the most significant dangers of contemporary algorithmic governance. The logic of security possesses an internal dynamic of expansion. The more the capacity for data collection and prediction increases, the stronger becomes the tendency toward continuous surveillance, constant classification, and preventive intervention. Technology may thus reinforce forms of authority much broader than those ever available to traditional states.
Precisely in opposition to this prospect, the liberal tradition of Locke acquires particular significance. Whereas Hobbes places security at the centre, Locke founds the political community upon the protection of the natural rights of the human being. State authority does not exist for itself nor is it legitimized exclusively by its effectiveness. It exists in order to safeguard the freedom, property, and personal autonomy of citizens. Sovereignty is henceforth subject to substantive limitations.
The significance of this theory for artificial intelligence is exceptionally great. Most contemporary issues concerning the protection of personal data, digital privacy, the use of biometric information, automated administrative decision-making, and the continuous monitoring of citizens can be regarded as new forms of the old liberal problematique. The basic question remains unchanged: up to what point is the expansion of state or private authority permitted without affecting the core of human freedom?
Artificial intelligence makes this question even more complex. Data collection now takes place in ways that often remain invisible to the citizen himself. Risk assessments, automated classifications, and mechanisms of individualized prediction can substantially affect his life without there always being full awareness of their functioning. The liberal demand for the limitation of authority therefore acquires a new dimension. It concerns not only traditional state interventions but also the complex digital infrastructures that daily affect the exercise of fundamental rights.
The third great modern contribution comes from Rousseau. The theory of the general will transfers the discussion from the protection of individual rights to the democratic legitimation of the legal order itself. Law is truly binding only when it constitutes an expression of the political autonomy of citizens. Obedience to the law does not mean submission to an alien will but participation in a collective process of self-legislation.
This problematique acquires particular topicality in contemporary digital society. A large part of the rules that now affect the daily life of human beings is not produced exclusively by democratic legislative bodies. It is shaped through private digital platforms, mechanisms of automated evaluation, algorithmic recommendation systems, and technological infrastructures that function on a global scale. The question thus arises whether the political community continues to determine itself or whether a significant part of normative force is transferred to private technological centres of decision-making.
This development far exceeds the level of individual protection of rights. It touches the very core of democratic theory. If the most significant social choices are influenced by algorithmic systems whose design, functioning, and priorities are determined outside democratic procedures, then not only individual freedom but also collective political autonomy is called into question.
Modern thought therefore arrives at a new synthesis. Law appears simultaneously as a mechanism of security, as a guarantee of individual rights, and as an expression of democratic self-legislation. These three dimensions are not always in harmony. They often conflict with one another, creating tensions that continue to characterize contemporary legal orders as well. Artificial intelligence does not create these tensions; it makes them, however, more visible and more intense.
The historical experience of modernity thus leads to an essential conclusion. Technological progress cannot be evaluated exclusively on the basis of the increase of effectiveness or predictive capacity. It must be integrated into a much broader framework, within which the protection of freedom, the preservation of fundamental rights, the democratic legitimation of authority, and the capacity of citizens to remain genuine co-shapers of the legal order within which they live are jointly assessed.
The next historical period will attempt to answer these very problems through faith in the power of right reason. The Enlightenment will highlight the generality, clarity, and predictability of the law as the pre-eminent guarantees against arbitrariness, thereby creating the theoretical background upon which the idea of impersonal algorithmic rationality will later be built.
CHAPTER VI
The Enlightenment, Legal Formalism, and the Ideal of Impersonal Rationality
The Enlightenment constitutes one of the most decisive periods in the history of legal thought, because it radically changes not only the content of institutions but also the very methodology by which law is approached. If the medieval tradition placed at the centre the relationship of human law to a higher moral order, and modernity highlighted the concept of sovereignty and individual rights, the Enlightenment attempts to construct a legal order that will rest primarily upon the power of right reason. The authority of tradition, religious authority, or historical custom gradually recedes and is replaced by the requirement that every institution be capable of being justified by criteria of universal rationality.
This new view is not confined to philosophy. It decisively affects legislation, the judicial function, and the overall organization of the state. The clarity, generality, predictability, and publicity of the law are elevated to fundamental principles of the legal order. The arbitrariness of the ruler or of the judge is henceforth regarded as the greatest threat to the freedom of the citizen. The antidote to arbitrariness is not the personal virtue of the governor but the objective functioning of rules that are known in advance, are applied in a uniform manner, and bind everyone equally.
The significance of this historical transformation for contemporary artificial intelligence is exceptionally great. A large part of the argumentation in favour of algorithmic decision-making rests precisely upon Enlightenment ideals. It is argued that algorithms apply the same rules to everyone, limit personal prejudices, ensure consistency, increase predictability, and render state action more objective. The basic promise of artificial intelligence thus appears as the continuation of a long historical programme, which aims at replacing personal arbitrariness with impersonal rationality.
This historical continuity undoubtedly exists. The pursuit of the uniform application of law constitutes one of the most significant achievements of the Enlightenment and continues to constitute an essential precondition of every rule-of-law state. Equality before the law would be impossible if every judge or every administrative official applied different criteria depending on his or her personal views. The desire to limit subjectivity is therefore an entirely legitimate objective.
However, the Enlightenment itself contains elements that allow a much more complex evaluation of artificial intelligence. Rationality is never exhausted in the consistency of results. It also presupposes the possibility of understanding the procedures through which these results are produced. The public use of reason constitutes a fundamental principle of Enlightenment thought. The citizen is not called upon merely to accept decisions because they are effective. He is entitled to know the reasons that ground them, to examine them, to challenge them, and to participate in public dialogue concerning their correctness.
This dimension acquires exceptional significance when the so-called “black box systems” are examined. Many of the most advanced machine learning models are characterized by exceptionally high computational performance but at the same time by a limited possibility of interpreting their internal procedures. The final prediction may be exceptionally accurate, without it always being clear why the system was led to the specific decision.
From a purely technical point of view, this difficulty may be considered acceptable insofar as the result remains reliable. From a legal and political point of view, however, the problem is much deeper. The legal order is not interested only in the correctness of the result but also in the legality of the procedure through which it is produced. The possibility of public control constitutes an essential element of the rule of law. A power that cannot be explained becomes extremely difficult to control, even when it acts with the best intentions.
This point reveals an interesting internal tension within the Enlightenment programme itself. On the one hand, the Enlightenment seeks objectivity, generality, and consistency of rules. On the other hand, it requires publicity, transparency, and the possibility of accountability of authority. These two pursuits go hand in hand as long as the logic of decisions remains intelligible. When, however, technical complexity renders the understanding of the very decision-making procedure impossible, then a new form of contradiction appears. Exceptionally advanced technological rationality may prove politically anti-Enlightenment.
This observation does not constitute a mere theoretical concern. The use of algorithmic systems in justice, tax administration, recruitment, the assessment of creditworthiness, social welfare, and the management of migration flows already creates cases in which citizens are substantially affected by decisions whose logic they have difficulty understanding. If the citizen is unable to be informed for what reason he was assessed as high-risk, for what reason his application was rejected, or for what reason he was assigned a specific administrative treatment, then the very possibility of effective judicial protection is significantly restricted.
The historical experience of the Enlightenment reminds us that legality presupposes much more than mathematical consistency. It also requires the existence of institutions that allow the citizen to know, to control, and to challenge the exercise of authority. The public reasoning of decisions does not constitute an administrative formality but an essential guarantee of political freedom.
The same problematique is also connected with another significant aspect of Enlightenment thought, namely the universality of law. Law must be applied without discrimination and without preferential treatment of specific groups. This principle constitutes one of the most significant achievements of the modern constitutional tradition. Artificial intelligence promises that it can effectively serve this universality, because it applies the same computational patterns to everyone.
Reality, however, proves more complex. The historical development of algorithmic systems has demonstrated that even when the mathematical procedure is unified, the data upon which it is trained may incorporate historical inequalities, social prejudices, or institutional distortions. The uniform application of a distorted model does not produce equality but the generalized reproduction of inequality. In this way, it is revealed that the Enlightenment idea of the generality of law always presupposes also substantive equality of the conditions within which it is applied.
This development leads to a deeper understanding of the relationship between the Enlightenment and artificial intelligence. AI constitutes neither the natural completion nor the denial of the Enlightenment programme. On the contrary, it functions as a test of its very principles. It obliges us to re-examine what objectivity, neutrality, rationality, and equality truly mean when decisions are produced through exceptionally complex computational systems.
Thus, the Enlightenment does not offer only arguments in favour of the use of artificial intelligence. At the same time, it provides the most significant theoretical criteria for controlling its limits. The requirement of transparency, public accountability, the possibility of challenge, and institutional controllability continues to constitute a necessary precondition of every form of legal authority, regardless of whether it is exercised by human beings or supported by algorithmic systems.
This historical course will lead, during the nineteenth and twentieth centuries, to the development of legal positivism. The pursuit of systematicity and formal validity will now assume its fullest theoretical form, creating the framework within which the idea of the complete formalization and algorithmization of legal thought will later appear.
CHAPTER VII
Positivism, Validity, and Algorithmic Normativity
The historical evolution of legal thought during the nineteenth century is characterized by a profound transformation of its theoretical self-understanding. Whereas the Enlightenment had already highlighted the significance of rational legislation, the generality of law, and the impersonal functioning of institutions, legal positivism attempts to confer upon law even greater scientific autonomy. Its basic pursuit is to detach the theory of law from constant ethical, political, and metaphysical controversies and to organize it as an autonomous science, with its own object, its own methods, and its own criteria of validity.
This choice should not be considered a historical accident. It corresponds to the broader developments of modern science. Physics, mathematics, chemistry, and later the social sciences all seek to acquire strict methodology and a clear object. Within this scientific environment, legal science too attempts to form a space free from evaluative uncertainties, so that it may function with the greatest possible systematicity.
The first great expression of this tendency appears in the theory of John Austin. For Austin, law consists primarily in commands of the sovereign, accompanied by the threat of sanction. The validity of a rule does not depend on whether it is just but on whether it comes from the recognized political authority. The distinction between what is just and what is valid as law now acquires central significance. This theory seeks to ensure objectivity in the scientific description of the legal order, avoiding the transformation of every legal analysis into moral evaluation.
This fundamental distinction will deeply influence the entire subsequent theory of law and presents particular interest for contemporary artificial intelligence. A large part of the first applications of legal informatics rested precisely upon positivist presuppositions. Since law constitutes a set of valid rules, which can be organized into a logical system, the assumption seems reasonable that these rules can be represented in computational structures and applied through automated procedures. The first legal expert systems were built precisely upon this conception. Legal judgment was treated as a procedure of selecting and applying rules, which are connected with one another through logical relations.
This development is reinforced even more by the theory of Hans Kelsen. The “Pure Theory of Law” attempts to remove from the science of law every element that does not strictly belong to its normative object. Ethics, politics, sociology, and psychology are considered different scientific fields, which must not be confused with the description of the legal order. Law constitutes a system of rules organized hierarchically, in which every lower rule derives its validity from a higher rule up to the fundamental normative presupposition of the Grundnorm.
Kelsenian theory presents striking analogies with many forms of contemporary computational representation of law. The idea of a hierarchized structure of rules, which can be classified, connected, and applied with logical consistency, fits particularly well with the first generations of rule-based systems. The legal order appears as a well-organized normative system, whose relations can be described with great precision.
However, the very evolution of artificial intelligence very quickly revealed the limits of this approach. Most contemporary machine learning systems no longer function through the direct application of rules. They do not begin from a predetermined logical structure nor do they always apply explicit normative provisions. On the contrary, they extract patterns from vast volumes of data, create statistical models, and form predictions based on probabilities. Their logic is empirical and predictive rather than strictly normative.
This development creates a particularly interesting theoretical problem. If the functioning of a system is based mainly on statistical correlations and not on explicit rules, then its relationship to the traditional concept of legal validity becomes unclear. The system may decisively influence administration or even justice, without however being easily integrated into the positivist conception of a rule of law.
This problem appears particularly intensely in predictive models. A system may predict with great accuracy the probability of recidivism of a convicted person, the probability of tax evasion, or the risk of financial insolvency. These predictions may substantially influence the decisions of the competent authorities, without however themselves constituting legal rules. We thus find ourselves before a new form of normative influence, which is identified neither with the traditional rule of law nor with mere factual information.
The theory of Herbert Hart allows this new reality to be understood more deeply. Hart differentiates himself from Austin by maintaining that law is not merely a set of commands but a social practice. Rules exist because they are recognized and applied by the institutional actors of the legal order. Particular significance is acquired by the famous rule of recognition, that is, the set of social criteria through which the officials of the legal order ascertain which rules are valid.
This Hartian conception presents exceptional interest for the age of artificial intelligence. The critical question is no longer only whether algorithms produce results but also whether their results are gradually incorporated into the very practice of the legal order. The more judges, administrative authorities, and public organizations base their decisions on algorithmic assessments, the more these systems cease to constitute simple auxiliary tools and acquire actual normative influence.
This transformation does not necessarily take place through formal legislative recognition. It may occur gradually, through everyday administrative practice. An information system that is used for a series of years for the assessment of tax risks, for the allocation of social benefits, or for the prioritization of administrative cases may substantially influence the way the legal order functions even without possessing formal normative force. Thus a new form of “factual normativity” is created, which does not rest exclusively on legal rules but on the continuous institutional use of algorithmic procedures.
This development simultaneously highlights the limits of classical positivism as well. The distinction between valid law and factual social processes becomes increasingly difficult when the very processes of decision-production are shaped through complex digital infrastructures. The question is no longer only which rule is valid but also which mechanism guides its application in practice.
The historical course of legal positivism thus leads to a particularly significant conclusion for the contemporary theory of artificial intelligence. The formal validity of rules continues to constitute a necessary precondition of the legal order. It is no longer sufficient, however, to explain the manner in which contemporary forms of normative influence are produced. Algorithmic governance creates new levels of decision-making, in which statistical predictions, classifications, and automated assessments substantially shape the functioning of institutions without always being integrated into the traditional categories of legal theory.
The next great turning point in the history of legal thought will come from legal realism. Whereas positivism focused on the validity of rules, realism will transfer interest to the actual functioning of courts and to the behaviour of judges, opening a new theoretical horizon that presents particular relevance to contemporary predictive justice systems.
CHAPTER VIII
Legal Realism and Predictive Justice: From the Prediction of Judicial Behaviour to Algorithmic Forecasting
The history of legal thought rarely follows a straight-line course. Every theoretical achievement gives rise to new challenges, while every attempt to systematize the legal order provokes reactions that seek to bring back to the forefront elements that had been underestimated. This is exactly what happened during the first decades of the twentieth century. The great development of legal positivism and conceptual formalism created the impression that the judicial decision could be presented as the result of an almost mechanical application of predetermined rules. This image was radically challenged by the movement of legal realism, which attempted to transfer interest from the abstract structure of rules to the actual functioning of the courts.
This shift was profoundly revolutionary. The realists did not deny that rules exist or that the legal order possesses a systematic structure. They maintained, however, that the understanding of law presupposes the study of the way in which rules are applied in practice. Interest is thus transferred from the text of the law to the judicial decision, from abstract normativity to concrete institutional behaviour. Law ceases to be regarded exclusively as a system of rules and is approached as a living social process.
Perhaps the best-known formulation of this new perspective comes from Oliver Wendell Holmes, according to whom law must be seen from the perspective of the person interested in predicting what the courts are going to do. The famous concept of “prediction” does not aim at reducing law to a statistical phenomenon. It aims at highlighting the reality that the legal order acquires practical meaning only through the institutional decisions that apply the rules to social life.
This position presents remarkable topicality in the contemporary age of artificial intelligence. Predictive justice systems rest precisely upon the idea that the processing of a large number of previous judicial decisions allows the extraction of patterns capable of predicting, with relative probability, the outcome of future disputes. Technology thus appears to realize, by computational means, what Holmes had described at a theoretical level more than a century earlier.
This similarity, however impressive it may be, must not conceal the essential differences between the two approaches. Holmes never maintained that prediction constitutes law itself or that justice is exhausted in the statistical regularity of judicial decisions. The concept of prediction functioned as an analytical tool, as a means of understanding the actual functioning of institutions. Contemporary artificial intelligence, by contrast, creates the temptation to transform prediction into a normative guide of the judicial function itself. It is precisely here that one of the most significant theoretical shifts of the digital age begins.
Historical experience teaches that there is an essential difference between the description of a phenomenon and its transformation into a standard of behaviour. Legal realism described judicial practice in order to understand it. Predictive systems risk using past practice as the basis for shaping future practice. This transformation has profound consequences. When a prediction influences the very decision that is to be made, a mechanism of self-confirmation is created, within which the past acquires increasing power to shape the future.
This problem becomes particularly evident when cases are examined in which previous decisions reflect historical inequalities or prejudices. A system trained exclusively on past practices may reproduce the same structures of inequality, not because it is programmed to act unjustly but because it treats the past as a reliable basis for predicting the future. Historical regularity is thus transformed into technological necessity.
This development highlights a deep philosophical difficulty. Law does not exist only to reproduce social reality but also to transform it. The great historical turning points in the protection of human rights, gender equality, the prohibition of racial discrimination, or the protection of socially vulnerable groups did not arise because they faithfully reflected previous practice. On the contrary, many times they constituted a conscious rupture with established forms of injustice. A system that draws exclusively from the historical past has difficulty recognizing the legitimacy of this creative transcendence.
The same problem also appears in judicial argumentation. The human judge does not evaluate only the probability of a solution but also its compatibility with the principles of the legal order, with constitutional values, and with the evolutionary course of law. Judicial judgment always possesses a creative dimension. It is not limited to the prediction of what happened in the past but examines which solution is justified under the specific circumstances of the present case.
This dimension acquires particular significance when the so-called hard cases are examined. The more a dispute deviates from ordinary cases, the less useful statistical regularity becomes. The historical uniqueness of the specific case cannot be mechanically extracted from previous data. It requires interpretation, balancing, and creative institutional judgment. Legal realism never denied this reality; on the contrary, it attempted to highlight it against the formalism of its time.
Contemporary artificial intelligence therefore brings back the same dilemma at a new level. On the one hand, the capacity to process millions of judicial decisions provides valuable tools for the study of case law, the detection of tendencies, and the strengthening of legal certainty. On the other hand, excessive trust in the predictive power of algorithms may lead to the gradual replacement of creative judicial judgment by the statistical reproduction of the past.
The history of legal thought shows that law has never evolved through the simple repetition of already known solutions. Every great transformation was born from the capacity of judges, legislators, and theorists to transcend creatively established practices when these proved inadequate. Predictive analysis may constitute a valuable scientific tool. It cannot, however, be transformed into the sole foundation of judicial judgment without altering the very historical mission of law as an institution of continuous renewal of justice.
With this observation, the contribution of legal realism to the diachronic evolution of legal thought is completed. The next great theoretical turning point will come from the interpretative theory of Ronald Dworkin, which will demonstrate that law is not only a system of rules nor a simple prediction of judicial behaviour, but above all a unified practice of principles and normative justification. This contribution proves decisive for understanding the limits of artificial intelligence in the so-called “hard cases,” where statistical prediction is not sufficient for the production of justice.
CHAPTER IX
Ronald Dworkin, Principles, and the Limits of Algorithmic Judgment in Hard Cases
The history of legal thought during the twentieth century is completed neither with positivism nor with legal realism. These two great theoretical traditions highlighted critical dimensions of law, but at the same time left important questions open. Positivism gave priority to the validity of rules, without sufficiently explaining the way in which judges resolve cases in which the rules are unclear, contradictory, or incomplete. Legal realism, on the other hand, turned attention to the actual functioning of courts, without however formulating a comprehensive theory concerning the normative justification of decisions. The theoretical synthesis of these problematics is attempted in the work of Ronald Dworkin, which constitutes one of the most significant interventions in contemporary philosophy of law and presents exceptional significance for understanding the limits of artificial intelligence.
The starting point of Dworkinian theory lies in the rejection of the conception that law is identified exclusively with a set of rules. According to Dworkin, every legal order includes not only explicit rules but also principles, which possess normative force even when they have not been expressly formulated by the legislator. These principles do not function like rules. They are not applied in the manner of “all or nothing,” but possess different weight and require balancing depending on the circumstances of each concrete case.
This distinction constitutes a turning point in the history of legal theory. Until then, a large part of the discussion revolved around the question of which rules are valid and how they are applied. Dworkin shifts interest toward the process through which the legal order seeks the best possible justification of its own decisions. Law ceases to appear as a static set of provisions and is revealed as a continuous interpretative practice, within which every new decision is called upon to be integrated harmoniously into the overall moral and institutional structure of the legal order.
This theory acquires particular significance for contemporary artificial intelligence, because it reveals a dimension of the judicial function that is difficult to render in purely computational terms. Most contemporary algorithmic systems achieve exceptional performance when they are called upon to recognize patterns, classify information, search case law, or predict possible outcomes of cases. Their functioning is based on the recognition of relations within very large volumes of data. Where, however, the legal order requires the balancing of fundamental principles, the situation changes substantially.
The so-called hard cases constitute a characteristic example. These are cases in which there is no clear legislative answer or in which more rules and more principles appear to lead to different results. In such circumstances, the judge cannot be confined to the mechanical application of predetermined provisions. He must examine the overall character of the legal order, balance competing values, evaluate the institutional coherence of the solution, and arrive at a decision which not only resolves the specific dispute but at the same time reinforces the coherence of law as a unified normative system.
This function differs radically from simple statistical prediction. A system of artificial intelligence can identify that in similar cases the courts have ruled in a specific manner. This information is certainly useful. It does not, however, answer the normative question whether the same solution still constitutes the most just one in light of the particular circumstances of the new case or of the evolution of the institutional principles of the legal order. The distance between prediction and justification becomes here entirely evident.
The Dworkinian concept of “integrity” makes the problem even clearer. The judge does not function as a simple applier of isolated rules nor as a statistical analyst of previous decisions. He is called upon to treat law as a unified system of principles, within which every new decision must be integrated in a manner that preserves the coherence of the legal order as a whole. Integrity does not constitute simple consistency between decisions. It constitutes internal harmony among institutions, values, and interpretative choices.
This requirement presents particular significance for artificial intelligence. Large language models and advanced machine learning systems are able to reproduce, with impressive success, existing arguments and to synthesize persuasive legal analyses. This capacity, however impressive it may be, is not necessarily identical with normative integrity. The production of a persuasive argument does not entail that this argument constitutes the best possible institutional justification of the specific decision.
This difference is connected with a deeper philosophical distinction between explanation and justification. Artificial intelligence can explain why a specific result appears more probable on the basis of the available data. Judicial judgment, however, requires much more. It requires that it be shown why the specific solution must be considered correct from a normative point of view, even when it deviates from previous practices or even when it overturns established conceptions.
The history of law offers numerous examples of such creative transcendences. Significant decisions concerning equality, human dignity, the protection of minorities, or the evolution of fundamental rights did not arise because they constituted the statistically most probable solution. On the contrary, many times they constituted a conscious revision of practices that had until then been dominant. This development would be extremely difficult to predict exclusively through analysis of the past.
This observation does not diminish the value of artificial intelligence in legal practice. On the contrary, it allows the more precise determination of its role. AI can constitute an exceptionally effective tool for mapping case law, detecting interpretative tendencies, comparing judicial decisions, and supporting legal research. It can significantly enhance the quality of the preparation of a judicial judgment. What it cannot fully replace, however, is the final act of normative responsibility, in which the judge personally assumes the obligation to justify publicly why a specific solution is the most just under the given circumstances.
The historical evolution of legal thought therefore shows that the problem of artificial intelligence is not primarily technical. It is a problem of legal theory. The more law is conceived as the mechanical application of rules, the more easily its full automation seems possible. But the more law is understood as an interpretative practice of principles, institutional coherence, and public justification, the more it becomes evident that human judgment continues to play an irreplaceable role.
Dworkin’s contribution thus marks a new culmination of the diachronic evolution of legal thought. The legal order no longer appears either as a simple set of rules or as a statistical regularity or as the product of pure political will. It is revealed as a dynamic system of principles, whose unity is preserved through the continuous interpretative effort of its institutional organs. Precisely this dimension will constitute the starting point of contemporary critical theories, which will challenge even the very concept of the neutrality of law, opening the discussion on algorithmic bias and the new forms of digital power.
CHAPTER X
Critical Theories, Algorithmic Bias, and the Deconstruction of Neutrality
The historical evolution of legal thought during the second half of the twentieth century is characterized by an even deeper shift. Whereas positivism sought to found the objectivity of law through the formal validity of rules, and Dworkin brought back to the forefront the significance of principles and interpretative coherence, critical theories challenged the very axiom of the neutrality of law. The question is no longer only whether a decision is formally valid or interpretatively justified. It now becomes necessary to investigate whether the very structures of law reproduce relations of social power, exclusion, and inequality, even when they appear to be absolutely neutral.
This transformation constitutes perhaps one of the most significant theoretical developments for understanding artificial intelligence. Most contemporary discussions surrounding algorithmic bias could not have been formulated without the theoretical legacy of critical legal studies, feminist theories of law, theories concerning racial equality, and sociological analyses of power. All these approaches converge upon one basic observation: neutrality is often an ideological form that conceals the actual relations of power operating within society.
This observation does not concern law exclusively. It concerns every institution that organizes social relations. Rules, procedures, administrative mechanisms, and techniques of evaluation often appear as objective, while in reality they reflect historical choices, cultural assumptions, and social hierarchizations. The history of law repeatedly proves that institutions which for centuries were considered entirely natural and neutral were later revealed to be mechanisms for maintaining inequalities.
The same historical experience is repeated today in the field of artificial intelligence. During the first years of the algorithmic revolution, the conception widely prevailed that computers, because they function mathematically, cannot be biased. The mathematical structure of algorithms was regarded as a guarantee of objectivity. Very soon, however, the practical application of these systems proved that reality is much more complex.
Bias does not necessarily appear at the level of the algorithm itself. It may already be incorporated in the training data, in the categories that are used, in the variables that are selected, in the labels assigned to examples, or even in the very definition of the problem that the system attempts to solve. Each of these stages constitutes a human choice. None of these choices is absolutely neutral.
This observation radically changes the way in which artificial intelligence must be evaluated. The problem is no longer confined to the technical accuracy of the models. Even a system with exceptionally high predictive success may reproduce existing social inequalities if it has been trained on data that record unequal historical practices. Mathematical consistency does not eliminate social bias; on the contrary, it may give it greater stability and a broader scale of application.
This element renders algorithmic bias a historical and not exclusively technological phenomenon. Algorithms do not create most forms of discrimination from the beginning. They often function as mechanisms for condensing and reproducing long-term historical developments. The social inequalities that were created over decades or even over centuries acquire a new form through the statistical patterns detected by machine learning systems. Technology does not necessarily invent new discriminations; it makes it possible, however, for them to be reproduced with unprecedented speed, accuracy, and extent.
This historical dimension explains why addressing algorithmic bias cannot be confined to purely technical interventions. The improvement of datasets, the balancing of samples, or the development of new methods of statistical correction certainly constitute significant steps. But they are not sufficient by themselves. If the very social relations that produce the data continue to be characterized by inequalities, then technical correction addresses only the symptoms and not the deeper causes of the problem.
This observation leads to a broader theoretical consequence. Artificial intelligence cannot be an object exclusively of computer science or engineering. Its understanding requires the cooperation of legal science, sociology, political theory, philosophy, history, and ethics. Algorithms always function within institutions, and institutions always function within historically formed societies. The detachment of technology from its historical and social context inevitably leads to an incomplete understanding of its consequences.
The significance of this observation becomes particularly apparent in the field of public administration and justice. When an algorithmic system is used for risk assessment, the allocation of social benefits, the selection of candidates, or the support of judicial decisions, it does not function in a social vacuum. Its decisions affect real human beings, who are situated within specific historical, economic, and cultural conditions. The formal equality of the computational procedure does not always entail substantive equality of outcomes.
Critical theories also highlight another particularly significant element, which is often overlooked in public discourse. The very selection of the problems assigned to artificial intelligence is already a political and normative decision. It is not neutral that enormous resources are invested in the development of systems of predictive policing, assessment of creditworthiness, or automated surveillance, while much less emphasis is given to applications that strengthen access to justice, the protection of socially vulnerable groups, or the transparency of public administration. The very direction of technological development reflects social priorities and relations of power.
The history of legal thought thus leads to a particularly significant conclusion. Neutrality can no longer be considered a given characteristic either of law or of artificial intelligence. On the contrary, it constitutes a continuous institutional desideratum, which requires constant public control, critical evaluation, and institutional guarantees. Objectivity is not a property automatically possessed by technical procedures. It is the result of complex institutional choices, which must remain transparent and open to democratic challenge.
With the highlighting of algorithmic bias, another significant stage of the diachronic evolution of legal thought is completed. Interest now shifts toward two fundamental concepts that acquire particular significance in the digital age: responsibility and accountability. For the more decisions are produced through complex technological systems, the more pressing becomes the question who ultimately bears responsibility for their consequences and in what manner the structure of legal responsibility can be preserved in an environment where human and algorithmic action are constantly intertwined.
CHAPTER XI
Responsibility, Personhood, Reasoning, and Accountability in the Age of Artificial Intelligence
The history of legal thought proves that few concepts have been as stable as the concept of responsibility. From the first forms of normativity to the most contemporary theories concerning human rights, the legal order is always organized around the idea that every act can be attributed to a specific bearer, that every decision derives from a person or institution possessing recognizable competence, and that every exercise of authority is accompanied by an obligation of reasoning and the possibility of accountability. This diachronic stability is not an accidental characteristic of the legal tradition. It constitutes one of the fundamental preconditions of the rule of law itself.
The emergence of artificial intelligence creates, perhaps more than any other previous technological development, a deep challenge to this historical structure. Algorithmic decisions are not always produced through directly recognizable human judgment. On the contrary, they are often the product of interaction among programmers, organizations, data providers, machine learning procedures, continuous software updates, and complex computational infrastructures. The traditional image of the legal relationship, within which a specific bearer decides and personally assumes responsibility for his decision, becomes increasingly difficult to apply without adaptations.
This transformation does not concern only civil or criminal liability for harmful acts. It concerns the theory of law as a whole. The very concept of normative attribution is placed under pressure. When an administrative decision rests substantially upon an algorithmic assessment, when a judicial judgment is influenced by a predictive justice system, or when an economic transaction is rejected because of an automated risk calculation, the question arises to whom exactly the final decision must be attributed. Is the programmer responsible? The organization that uses the system? The entity that trained it? The user who activated it? Or do all participate in a different way in the formation of the result?
This difficulty is not exclusively practical. It reveals a deeper historical shift. Classical legal thought was built around the concept of the person as a bearer of rights, obligations, and responsibility. From Roman law to contemporary constitutional theory, the person constitutes the point of reference of the legal order. Artificial intelligence, however, functions through distributed procedures, in which the production of a result cannot always be attributed to a single will. This historical development creates a tension between the structure of digital systems and the structure of legal responsibility.
The answer to this problem cannot be sought in the recognition of legal personality in artificial intelligence systems themselves. Although relevant proposals have been formulated from time to time, the diachronic evolution of legal thought leads in a different direction. Responsibility is not connected only with the production of results but also with the moral and institutional possibility of accountability. The person is not merely a bearer of actions; he is a bearer of reason. He can explain, justify, change his stance, assume the consequences of his acts, and be subjected to public control. Artificial intelligence, however advanced it may be, does not possess this form of normative self-consciousness.
Precisely for this reason, the history of law insists upon the preservation of human responsibility even when exceptionally complex technological means are used. The use of tools does not remove the accountability of the one who decides to use them. This principle appears already in the first forms of private law and is preserved up to contemporary theories of state liability. Technology may change the manner of decision-making; it cannot, however, eliminate the need for the existence of a responsible bearer of the decision.
Particular significance is acquired in this context by the concept of reasoning. The requirement of a reasoned decision runs through the entire history of the European legal tradition. From the Roman ratio decidendi to contemporary constitutional guarantees, reasoning is not considered a decorative element of the judicial or administrative function. It constitutes an essential precondition of legality. Authority is justified because it can explain the reasons for its choices.
The significance of this principle becomes even greater in the age of artificial intelligence. So-called explainability is not merely a technical property of an information system. It constitutes the contemporary expression of a historical legal principle, according to which every person who suffers legal consequences is entitled to know why the specific decision was taken. This right is directly connected with the right to be heard, with effective judicial protection, with the principle of equality, and with the public control of authority.
If a decision cannot be explained, then its challenge also becomes exceptionally difficult. The interested person does not know which element of his data was considered critical, which criterion was evaluated as more significant, or which logic led to the final result. Without knowledge of the reasons for the decision, the possibility of effectively contesting it is drastically restricted. In this way, the absence of reasoning is not only a problem of transparency but also a problem of judicial protection.
Historical experience shows that every form of authority that detaches itself from the obligation of reasoning tends also to distance itself from democratic accountability. The same is true of algorithmic systems. The more their functioning remains opaque, the more difficult the substantive control of their influence upon the legal order becomes. Technological complexity cannot constitute a reason for exemption from the legal obligation of reasoning; on the contrary, it renders this requirement even more imperative.
The diachronic evolution of legal thought thus leads to a clear theoretical position. Artificial intelligence can transform decision-making procedures, but it cannot transform the fundamental principles of accountability. Responsibility continues to presuppose a human bearer, legality continues to require reasoning, and the democratic polity continues to presuppose the possibility of public control of every form of authority. Technology must adapt to these principles, and not the principles retreat before technological complexity.
This problematique naturally leads to the final great stage of the diachronic evolution of legal thought. Contemporary theory of human rights attempts to construct an overall normative framework within which artificial intelligence may develop without the fundamental values of the rule of law, democracy, and human dignity being put at risk.
CHAPTER XII
Human Rights and the Contemporary Normative Limits of Artificial Intelligence
The historical course examined in the preceding chapters inevitably leads to the contemporary age of human rights. If ancient Greek thought formulated the first philosophical questions concerning justice, Roman science shaped the systematic technique of law, medieval theory highlighted the common good and natural law, modernity founded state sovereignty and individual rights, the Enlightenment advanced reason and the generality of law, positivism elaborated the concept of validity, realism highlighted the significance of the actual functioning of institutions, while interpretative and critical theories brought principles, justification, and relations of power back to the centre, then the theory of human rights appears as the point at which all these historical paths converge and acquire a new unity.
The contemporary protection of human rights does not constitute yet another special category of rules of positive law. It represents a new conception of the very structure of the legal order. The human being is no longer treated only as a subject of law nor only as a citizen of a specific state. He is recognized as a bearer of inviolable value, which precedes every form of political, administrative, or technological organization. This development also radically changes the way in which artificial intelligence is evaluated.
Public discussion surrounding AI often focuses on effectiveness, innovation, or economic development. All these dimensions are certainly important. The history of legal thought, however, reminds us that technological progress cannot constitute an autonomous criterion of legitimation. Technology acquires normative legitimation only when it is integrated into an institutional framework that protects the human being as an end and not as a mere object of data processing.
Precisely for this reason, the language of human rights today occupies a central position in almost all international and European normative texts concerning artificial intelligence. The protection of private life, human dignity, equality, the prohibition of discrimination, the protection of personal data, effective judicial protection, freedom of expression, freedom of thought, and democratic participation no longer appear as parallel limitations of technology. They constitute the very preconditions of its lawful use.
This development presents particular historical interest. The theory of human rights functions in a certain way as a contemporary form of natural law, without however being identified with classical natural law theories. Where Aquinas sought the higher measure of justice in natural law and right reason, the contemporary legal order seeks it in the universal protection of human dignity and fundamental rights. The historical continuity is evident, although the philosophical vocabulary has changed.
Artificial intelligence renders this development even more necessary. Advanced computational systems possess an unprecedented capacity to collect, combine, and analyse information. This capacity creates new forms of power, which far exceed the possibilities of traditional administrative mechanisms. Continuous monitoring, predictive analysis of behaviours, automated classification of citizens, the creation of profiles, and the extraction of conclusions about their personal choices substantially change the balance between the individual and authority.
The protection of privacy thus acquires new content. It no longer concerns only the prevention of unlawful disclosure of information. It concerns the very capacity of the human being to freely shape his personality without being under continuous computational assessment. Constant algorithmic monitoring does not affect only external behaviour. It may gradually change also the internal self-understanding of the human being, creating an environment within which freedom is limited not through direct coercion but through continuous digital evaluation.
Analogous significance is also acquired by the principle of equality. The history of legal thought repeatedly demonstrated that formal equal treatment is not always sufficient for the realization of justice. Artificial intelligence confirms this observation in a new way. An algorithmic system may apply an entirely uniform procedure to all citizens and at the same time produce systematically adverse results for certain social groups, because historical inequalities have already been incorporated into its training data. Real equality therefore requires continuous evaluation not only of the procedure but also of its outcomes.
The concept of human oversight also acquires new significance. It is often presented as a technical safety requirement, according to which a human being must be able to intervene when the system presents an error. The history of legal thought shows that its function is much deeper. Human oversight constitutes an expression of the principle that final normative responsibility cannot be detached from the human person. Even when artificial intelligence participates substantially in the decision-making process, the final judgment must remain integrated into institutions of human accountability.
The same is true of the possibility of challenging algorithmic decisions. The right of recourse before an independent court or other competent body does not constitute a simple procedural guarantee. It constitutes a fundamental precondition of democratic legality. No form of authority can be considered compatible with the rule of law if its decisions cannot be reviewed and revised.
The contemporary theory of human rights therefore leads to an overall revision of the way in which artificial intelligence must be evaluated. The basic question is not whether algorithms are more or less effective than the human being. The essential question is whether their use preserves unchanged the fundamental principles that constitute the democratic legal order. Effectiveness is certainly an important good. It does not, however, constitute the supreme good of law.
From this perspective, it becomes evident that artificial intelligence does not merely introduce new technical issues. It obliges legal science to re-examine the relationship between technology and anthropology, between effectiveness and justice, between authority and dignity. The history of legal thought reveals that every great historical period had to redefine this balance. Our own age is no exception.
With the completion of the examination of human rights, we reach the final stage of the present study. What remains now is to synthesize the overall picture of the diachronic evolution of legal thought and to evaluate definitively the place of artificial intelligence within it, in order to ascertain whether AI constitutes a rupture with the history of law or, on the contrary, the newest chapter of a history that continues to revolve around the same diachronic questions concerning justice, freedom, and human responsibility.
CHAPTER XIII
General Conclusions – Artificial Intelligence as a New Phase in the Diachronic Evolution of Legal Thought
The diachronic review that preceded makes it evident that artificial intelligence cannot be adequately understood as a simple technological achievement nor as yet another new object of legal regulation. Its significance far exceeds the limits of computer science, engineering, or even positive legislation. Artificial intelligence constitutes a profoundly theoretical fact, because it brings back to the centre almost all the fundamental questions that have shaped the history of legal thought from antiquity to the present day.
The examination of the early forms of normativity demonstrated that law was born as a technology for organizing social coexistence. Long before systematic legal theory appeared, human communities sought mechanisms for limiting violence, stabilizing social relations, and restoring collective peace. This function has not disappeared. On the contrary, artificial intelligence introduces new forms of social organization, in which the classification, prediction, and evaluation of behaviours acquire unprecedented extent. The original problem of normative organization reappears today through different technical means but with essentially similar theoretical content.
Greek philosophy was the first great moment in which law became an object of self-conscious reflection. The distinction between nature and law, the search for justice beyond enacted legality, the concept of equity, and Aristotelian theory of practical wisdom continue to constitute the most significant theoretical limits of every project of complete automation of judicial judgment. Artificial intelligence can support the logical application of rules; it has difficulty, however, reproducing that form of wisdom through which the judge recognizes the uniqueness of the concrete case and corrects the inevitable imperfections of the generality of the law.
Roman legal science taught that law is at the same time technique and interpretation. Conceptual systematization, logical precision, and methodological discipline constitute necessary preconditions of every mature legal order. Artificial intelligence makes use precisely of this centuries-long systematization of law. At the same time, however, the Roman tradition itself reminds us that the technique of law was never a mechanical process. Interpretatio always constituted a creative intellectual activity and not a simple application of predetermined schemes. This distinction continues to constitute the most significant limit of purely algorithmic legal thought.
The Christian and medieval tradition added yet another decisive dimension. Law does not exist only to organize social life effectively but also to serve a higher conception of justice, human dignity, and the common good. This observation acquires particular significance today, because public discussion surrounding artificial intelligence often tends to be confined to criteria of efficiency, accuracy, or economic utility. The history of legal thought reminds us that effectiveness is always a means and not the end of the legal order.
Modernity highlighted state sovereignty, natural rights, and democratic autonomy as new bases for the legitimation of political authority. In contemporary digital society, the same questions reappear in a new form. Who truly exercises authority when critical social functions are transferred to algorithmic infrastructures? Who controls the procedures through which decisions are made? How are fundamental rights protected when data processing takes place within global technological networks? These questions are not different from those that occupied modern political philosophy. They differ only as to the technological means through which they are manifested.
The Enlightenment introduced the idea of impersonal rationality, the generality of the law, and the publicity of authority. Artificial intelligence initially appears to constitute a natural continuation of this programme, since it promises greater consistency, less arbitrariness, and increased predictability. Yet the Enlightenment tradition itself also imposes its limits. No form of rationality can be considered compatible with the rule of law if it does not permit public control, transparency, and the possibility of understanding the procedures through which authority is exercised. The opacity of complex algorithmic models is not merely a technical difficulty; it constitutes a challenge to one of the fundamental principles of the modern democratic polity.
Legal positivism, realism, and Dworkin’s interpretative theory complete the overall picture. Positivism demonstrated the significance of normative validity. Realism reminded us that law exists mainly through the institutional acts of the courts. Dworkin highlighted that final judicial judgment is always an act of interpretative justification and not a simple application of rules or statistical patterns. Artificial intelligence simultaneously confirms all three of these theoretical traditions, but it cannot be fully identified with any of them. Its functioning exceeds the level of the formal application of rules, without however succeeding in completely replacing the creative interpretation that characterizes the judicial function.
Critical theories and the problematique of algorithmic bias demonstrated that neither law nor technology is ever absolutely neutral. Algorithms do not exist outside history. They are trained on historical data, function within specific institutions, and are influenced by social choices that precede their technical implementation. Addressing algorithmic bias therefore presupposes not only technical corrections but also deep historical and social self-knowledge.
The analysis of responsibility, reasoning, and accountability led to yet another critical conclusion. The legal order cannot function without the possibility of attributing responsibility to recognizable persons or institutions. Artificial intelligence may transform the manner of decision-making; it cannot, however, abolish the requirement that there exist a responsible bearer who assumes their public accountability. This principle constitutes a historical constant of the entire Western legal tradition.
Finally, the theory of human rights offers the normative framework within which all the preceding historical achievements can be synthesized. Human dignity, privacy, equality, freedom, judicial protection, democratic accountability, and respect for personality do not constitute obstacles to technological progress. They constitute the very preconditions of its legitimation. Artificial intelligence cannot be considered compatible with the rule of law because it is technologically advanced; it can be considered compatible only to the extent that it remains integrated into an institutional environment that safeguards the fundamental values of the legal order.
The overall historical review therefore permits the formulation of a final conclusion of particular theoretical significance. Artificial intelligence does not abolish the diachronic evolution of legal thought nor does it create an entirely new science of law. On the contrary, it renders the deep knowledge of the history of legal philosophy more necessary than ever. The more technological capabilities increase, the more it becomes evident that the decisive questions remain the same: what justice is, what the limits of authority are, how human dignity is protected, what the relationship between rule and judgment is, who is accountable for decisions, and what the place of the human being is within the legal order.
Perhaps the greatest lesson offered by the history of legal thought is that no technological revolution has ever managed to replace the fundamental normative principles of law. Institutions, forms of authority, techniques for applying rules, and means of organizing social life have changed. Yet the need has remained stable for law to constitute not merely a system of effective procedures but an expression of justice, freedom, and human responsibility.
Artificial intelligence is therefore not the end of the history of legal thought but its newest and perhaps most demanding chapter. The challenge of our age is not to decide whether the human being will be replaced by the machine. It is to ensure that technology will remain a tool of law and will not be transformed into an autonomous criterion of justice. Only if historical experience, philosophical foundation, and the principles of the rule of law continue to guide the development of artificial intelligence will it be possible for this new technology to serve the human being without undermining the very values upon which the legal order has been diachronically built.