Accueil RED&S

Syracuse University
Legal Expert Systems Conference

Applications : fields and conditions; advances and prospects; present limits to be considered and new boundaries to be explored, from technological, ethical and methaphysical viewpoints

Lucien MEHL
Conseiller d'Etat honoraire




The subject of this lecture is not exempt from risks which present antagonistic aspects. In the one hand, it can be considered that legal decision making can be and also may be computerized. In the other hand, the alleged precautions that seem to be taken about such a computerization could be seen as a pusillanimous attitude before the progresses of science and technology.

In the last decades, we observed an important development and improvement of the automated processes in the field of the private and public administrations. For a number of these applications of computer, the expertise in the art of managing and mastering the relevant knowledge and technics led to successfull issues.

Nevertheless,the situation is different when the matter consists no more, or not only, in automating administrative operations but working out decisions, more particularly if this purpose is not limited to the preparation of the decision to be taken, but includes the decision making stricto sensu , that is to say the final act of the decision, especially in the legal field. The case is less exacting, if the task assigned to the machine is not to enuntiate the complete and final decision, but only to propose alternative solutions, which is the function of the legal expert systems, qualified as "assistant" . In this assumption, the design of such a partial automation gives rise to a range of interrogations and raises up specific difficulties which may be of technical nature. But these generally are not of a major importance and, if it is the case, it is probable that they are to be surmounted in a relatively near future. The principal difficulties rather are of legal, ethical, political or even metaphysical nature, when legal automation concerns the whole decision making process.


It then appears that, in order to answer the question concerning the feasibleness and legitimacy of automating legal decision processes, it is necessary to wonder what is a decision, word in common use, but the bearing of which is not obvious. This will be further examined in the first part of this introduction.

The reason of these oratorical precautions resides on the fact that the recourse to systems science, computer science, new information technologies and, more generally, to artificial intelligence in legal activities is still controverted. Not so long ago, the resistance against or, at least, the reluctance about the so-called legal automation was very strong. It has now diminished, but it has not disappeared. It is probably a question of distrust or uneasiness, in regard, not only towards machines, but also towards, the so-called "scientism" or "technocracy". But we also have to consider what, in these attitudes of suspicion comprises a part of legitimacy. It seems

that the legal activity, especially the higher legal decision, cannot and may not only result from an automatic computation.

In the first part of this lecture are considered the types of decision of legal character, that may be deemed as computerizable, from the view point of the logic of the decision, as well as from the technical level.

But the main question, studied in the second part, is to examine if it does exist decisions that are not, or, at least, not fully computerizable and what are the reasons of the obstacles. Is it because this automation should be technically impossible or overweening? Or, is this computerization, in principle technically possible, prohibited for legal, ethical or metaphysical reasons, except if the intervention of a human decision maker is foreseen at the beginning, in the course or at the end of the decision process ?

Of course, it is indispensable to consider the present state of science, technology and law, in connection with the applications of legal automation. The advances attained in this last decade of this century shall be envisaged. Finally the present and foreseeable limits of legal automation and the new boundaries to be explored, in the field of technology and from a legal and philosophical viewpoint will be the matter of the third and last part of this lecture.


In the frame of this introduction the concept of decision will be briefly considered on a general plan. The matter in hand is the decisions that are taken by business undertakings, political or public administrative authorities and also by courts. This last category of decisions presents important and original aspects. For sake of briefness, the decisions taken by individuals in the management ot their private affairs are not examined here, in spite of the interest of this question.

One may consider that the concept of decision is, at least intuitively, clear. A decision is taken in order that a result sould be obtained, that an issue to a situation might be found or, at least, that the way to be followed could be chosen.The decision is often preceded by a preparation during which the required information is gathered and processed and then confronted with objectives and constraints. This stage of preparation is followed by that of the deliberation which consists in weighing the different issues that can or may be envisaged. Finally, an issue is chosen.

Of course, such a choice occurs if there is something to be chosen, which is not always the case. This assertion could be considered as a truism, but it is not so obvious and, in practice, often not sufficiently taken into consideration.

For a choice supposes that its author, who comes to the final act of the decision process, disposes of an amount of latitude, of liberty, in presence of a certain number (two or more) of convenient or, at least, acceptable issues. If the issue that can or may be adopted is the only one, this final act is not a genuine decision, nor its author really a decision-maker, even though this final act should be regarded as important, especially in its consequences. Anyway, the final stage of the process, especially if it concerns an authentical decision closes a sequence of phases (possibly with going back). This notion of closure appears in the etymology of the words decision in different languages, for exemple, Entschluss in German, beslissing in Dutch, beslutning in Norwegian. The English and French word is bound to the notion of cutting (latin, caedere) as well as the Russian word (rechenie, from rezat', to cut). One may indeed observe that deciding is, in a certain sense, to cut in the complexity of the representation of the reality and in the reality itself (to cut "the gordian knot " ). Less incisively - it is the case to say so - the notion of decision suggests, in principle, that a choice is exercised.

This question, concerning the existence of an option, is the criterium which enables to characterize the " true " decision. What then can be said, when the whole situation is determinate, so that there is no choice to be exercised ? The answer is that, in such a situation, there is not, actually, a decision to be made, even if the act in consideration is nevertheles denominated as such, which is however frequent, especially in the field of public administration. This kind of acts could be called pseudo-decisions, this term involving no pejorative connotation. Since the word "decision" is ambiguous, it is advisable to analyse the deep structure of the true (or genuine) decision which is not immediately apparent. Without excess of reductionism, it seems possible to propose a ternary structure of the decision, which has a descriptive and explanatory bearing, especially when the computerization concerning decision-making is to be envisaged.


In this ternary structure, one may, essentially, distinguish three parts, that it to say : the calculable part, which is, of course, computerizable; the uncertain part (or part of uncertainty), which is , to a certain extent, rebellious to calculation; the axiological part, which concerns ethical, cultural, political and even metaphysical values, hence partly legal at least. The appreciation of these values and the comparison between them are, in principle, but non entirely, excluded from possibilities of calculation. This is due to the complexity of the values, considered from an individual and, above all, social viewpoint. In practice, one of these parts of the decision can have a secondary importance or even be inexisting. For example, if the issue to be prepared and then adopted is wholly calculable, this issue is a a pseudo-decision and its preparation does not rise questions concerning uncertainty or axiology. It can also arrive that the issue should be quite obvious and immediate, so that no calculation is required.The importance or the existence of these three sections of the decision are very variable, according to the nature of the situations to be considered or of the acts, designated as decisions rightfully or by abuse of language.

The precise meanings of the epithets calculable, uncertain, axiological now are to be explained and developped.

1. The adjective calculable is here taken in a broad sense. It concerns, not only what is mathematically calculable, but also the elements of the decision for which a logico-deductive process is, wholly or partly, fully relevant. Such a process is a sort of calculation, even though it is expressed in "natural language" and not, or only partly, in symbolic notations. To the phrase "calculable part", might be prefered "zone of the determinate".

If the senses of the words calculable or determinate are obvious, it is however necessary to search into the extensions and limits of the calculable, especially in the situations, relatively frequent, presenting hazards or risks, the processing of which comes under the probability theory. This will be examined in the first part of this contribution concerning the acts that are entirely calculable (pseudo-decisions), since the aleatory situations or events are calculable, at least partly, which is not the case for the situations or events which are uncertain stricto sensu , that is to say when the required information is totally or highly lacunary.

2. This uncertain part of the decision is now to be investigated. What is uncertain stricto sensu , not calculable or, at least, only very partly calculable ? If probabilistic methods may eliminate a part of uncertainty, this favourable situation is rather exceptional. Nevertheless, if a decision, in a situation of uncertainty, must be made, especially in case of emergency, the decision maker is obliged to assume risks, which can be only calculated to a certain extent or level. But this partial calculation lets subsist shadow areas. In such cases, the decision, individual or collective, comprises an element of wager, in the non futile sense of this word.

3. It now remains to consider what concerns axiology. By that is understood what has reference to values, in the philosophical sense of this word, not in the economic sense, for the economic values are, in principle calculable, even if they can be imprecise or subject to difficulties of valuation which sometimes may be mastered through statistical or empirical methods. Nevertheless, the economic value may present, in certain situations, an axiological bearing, especially when it is connected with social aspects.

A value, in the philosophical sense, can be defined as a situation of socio-economic, socio-cultural or socio-political character, which is wishable to maintain, if it does exist, to preserve it, if it is threatened (or even to restore it) to attempt to reach it, if it is not yet the case.

Thus, in a given society, values are, in variable proportions for each of them, constraints or objectives. Very frequently, these values are legally defined. The importance of law and, more and more, of written law, is noticeable in our contemporaneous societies, so that law, more precisely the respect of law, the conformity with law, are a part of the socio-political values.

Since the philosophical values are not calculable, except partly or exceptionally, the decisions based on such values, require human intervention. The notion of values is explicited in the second part of this lecture, where are considered the acts that are non calculable or not entirely calculable, that is to say the genuine decisions. But is to describe forehand the acts entirely calculable or pseudo-decisions.



It is whith reference to the considerations that have preceded, about the structure of the decisions or acts designated as such, that it is posssible to operate a fundamental distinction about the feasibleness and legality of automation, between the decisons-choices, on one hand, and the pseudo-decisions, on the other. For this last category, their authors do not have at their disposal a latitude or margin of appreciation. The absence of such a margin implies that the act is fully calculable.


As it has been said before, the word "pseudo-decision", which has been proposed, does not convey a pejorative connotation ( in spite of its greek etymology : pseudomai, to lie) ; "pseudo-decision" is not more pejorative than "pseudonyme".

Thus, inside the set of the decisions and acts, often designated by the same word, is proposed a dichotomy between the decisions-choices (decisions stricto sensu ), which are true (or genuine) decisions, and the pseudo-decisions which can have the appearence of decisions, but the author of which only is an executive, non necessarily on a minor position, but who is not actually a decision maker.The pseudo-decision is the act that only results from a calculation, in the broad sense of this term, and consequently leads to a univocal result. The pseudo-decisions are such that two or more qualified persons, operating separately and applying the same rules, come to the same result.

In the precedent definition, the so-called "act" is a product of mental operations, then situated in the field of information, excluding consequently material changes. The act in question can be (but not necessarily) an "instrumentum juris ", a legal act, but it is not a genuine decision, in spite of its legal effects.

It is clear that, due to its characteristics, the pseudo-decision, most often, can be and may be computerized. This act, only secondarily, then comprises a lot of operations of the human mind. The computer, supposed to be duly programmed, does the essential and even the total of the logical and arithmetical calculation. The pseudo-decision is thus calculable, de facto and de jure . It is at least theoretically computerizable, which is frequently the case in practice. In such a situation, the expert system takes the place of the agent . From a legal viewpoint, the author of a pseudo-decision only has, in any case, a bound competence. Whatever be the process of his act, emanating from his own mental deduction or from the succession of the computer operations, the result will be the same. The automation of such acts, if conveniently conceived and led, cannot be prejudicial to the employee, citizen or undertaker who are concerned by them. Anyway, the interested parties may contest the act in question before the hierarchical authorities or before a court. It is to be added that if pseudo-decisions, in practice, generally are lawfull, it is because they proceed from non equivocal rules which, pre-established at a superior level, result from decisions stricto-sensu .


It has been supposed, up till now, that the logical and arithmetical calculation did not rise particular difficulties, even if it comprises an important number of fixed or variable data and even if it is highly complicated. This hypothesis seems to be acceptable since the duly programmed computer is fitted to master complication, notion that must be distinguished from complexity. A complex situation is not always suceptible to be mastered by a machine, since complexity can be defined as what cannot be analytically described.

The decision maker or his assistants sometimes have to face situations, the descriptions of which requires the recourse to probabilistic methods. In certain cases, in spite of the existence of aleatory elements, the issue could be entirely calculable, so that it presents a high level of certitude. Such is the case, when the "law of great numbers" is applicable or when the "game theory" leads to a sure conclusion. It also happens that information is insufficient and prevision doubtful or even impossible.

Then, it is required to take decision in a situation of incertitude. Such a decision is a true decision, since the decision maker has to exercise a choice. This choice is not necessarily of an axiological nature, though there are often ties between the axiological and the uncertain. The decision in uncertainty however does not fundamentally proceed from a preference. It is characterized by the fact that the decision maker, individual or collective, assumes risks and that he bets on the expected issue. In fact, it is rarely a pure and simple wager. That could be the case, if the decision maker has neither means or elements of information nor possibilities of prevision about the events to come or about the behaviour to envisage from the adversaries or the competitors. If some information elements can be found, the game theory may offer some assistance. If will be then a set or reckonings, even only partial, (in the sense of incomplete), whereas the calculation generally plays a minor rôle, if non existent, in the axiological field.

THE NON CALCULABLE ACTS (or not entirely calculable)

In this third part, are to be considered, the "true decisions" or decisions-choices. The presence of a possible part of the calculable in the decisions-choices will be examined.


As soon explained, what characterizes the decision-choice is the presence in its process and issue of an axiological part. The elements of this part, the values, are to be elucidated and weighed.The notion of value has been already defined and a distinction has been proposed between socio-economic and ethico-political values. One may assert that these values, even when they are not made explicit nor clearly defined, constitute a system, in the general sense of this word, but also in the sense it has in systems science.

In a system of values, those are, in principle, compatible. It is even possible that two values of the system mutually reenforce themselves : for example, dignity (respect of the human being) and liberty of a person, of a group, of a nation). But it may happen that values, although compatible, could be competing at a given level. Such is the case for security (individual or collective). Security, at a high level, necessarily leads to a reduction of liberty, but this reduction may be regarded as acceptable and even whishable, at least up to a certain point. In this brief description, it is excluded that the different value systems, conceivable or existing, will be analysed. It would be not so usefull as well, since what is in question is the non-calculability of the axiological, especially in the frame of the computerization of the legal decisions.

However, in order to avoid an excess of abstraction and, also, to make clear the non-calculability of the axiological part of the decision, it is useful to present a sketch of a value system that could be representative of what is existing or whished in occidental countries.

Of course, it is not the place here to dissert on values which concern, principally (but non exclusively), the human being as an individual like love, friendship, religious faith or political orientations. The individual propensities, feelings, inclinations, behaviours which derive from them are not be examined, since the main question, in this lecture, concerns computerization possibilities, especially from a legal and ethical viewpoint of the decisions that are prepared and taken in private or public institutions or organisms.

It is therefore advisable to precise what is understood by socio-economic and ethico-political values. The first set comprises security, well-being and knowledge. To be attained, these values generate a cost. The second set, dignity, liberty, justice have essentially moral bearings. Their cost has only a secondary aspect. These values have a moral basis and must be warranted by the political power, which explains they are called ethico-political.

This enumeration requires a brief explanation. By security is understood individual and collective security, in a broad sense. It comprises social security, which is an individual security confered by social institutions. The main elements of well-being are the standard of living and physical and mental health. In knowledge, are included general education and professionnal capacities. Dignity is all that concerns the respect of the human being. The value "liberty" is the liberty-autonomy, which shall be opposed to liberty-entropy, that, generally, engenders disorder. Finally, by justice is meant the virtue of justice (Gerichtigkeit , in German) which not only appears under its traditional aspects (neminem laedere, suum cuique tribuere) , but also with reference to the necessary principles of our time, like solidarity.

In order to take these values into consideration, in the decision process, the recourse to logico-deductive and, a fortiori , numerical calculation are not generally relevant, at least for the essential. This part of the decision process is thus incumbent on the appreciation of a human being, or of a group of human beings, duly entitled to do so by their legal status. In such a situation, the decision maker can be a judge, a tribunal, the qualified manager of a firm or of a public administration, a local government Council, a Parliament, a mayor, a high civil servant, a minister, a national or federal government, etc.. Whatever be the decision maker, individual or collective, private or public, or even the level of the decision (of course if it is a genuine one) the situation is the same, at least in principle as to the ethical imperatives, the difficulties of the choices to be made, and the liability of the author of the decision .


The non-calculability of decisions including appreciations of values however requires important observations and even reserves. Non-calculability does not signify irrationality. The axiological can partly be sustained by a mathematical calculation, especially when socio-economic values are concerned. As to the logico-deductive reasoning, it can efficiently intervene in the field of the ethico-political values. More generally, it is easy to observe that a number of decisions presenting an axiological preponderance also contains an authentically calculable sector, that it be mathematical or logico-deductive.

This can be examplified by the typical situation of the Dutch people whose a large part of their national territory is below the sea level. They are obliged to protect themselves by dikes against the fury of sea overflows and tide races. Unfortunately, the frequency, height and violence of an eventual overflow cannot exactly be foreseen. Taking into account the data of the past events of this nature, the Dutch authorities have built and they maintain in good conditions a number of dikes up to a certain height. But how to decide the required height ? If the people or his representatives exact an absolute protection, the cost for the nation shall be very high. If they adopt a lower degree of protection, the risk of inundation is greater and the cost of personal and material damages resulting from it can be superior than the cost of a stronger protection. Finally the representatives of the people, that is to say the Parliament, in cooperation with the Government, have to make a reasonable arbitration between the risk of a damaging inundation and the cost of a quasi-absolute protection, that is to say an arbitration between security and well-being, since a high protection is very expensive : it induces an increase of the public expenditures and, consequently, of the tax amounts and a reduction of the level of life. But what is the reasonable arbitration in such a situation ?

There is no model of calculation and reasoning, even with the help of a powerfull computer which could determine the terms of exchange between security and well-being, for they are heterogeneous values, without bridge between them. In this situation of "no bridge", the decision is only incumbent on man, on the people, or, as it has been said, on his representatives. Surely, for this type of decisions, a judicious use of meteorology, the recourse to historical statistical tables, technological improvements and so on, are of a major importance. But after these necessary steps, when the time has come to choice between levels of competing axiological values, the last word belongs to a human, to an individual or collective decision maker. The issue cannot emanate from a programme and the computer is impotent.

If the example issued from the situation of the Dutch people is higly significative, the problem of arbitration between axiological values is very general. Security and liberty are partly antagonistic. Dignity which comprises equality in rights is, of course, compatible with justice, like equality of opportunities. Solidarity also is in liaison with justice. It implies a certain amount of social redistribution which generates a reduction of income inequalities. But egalitarism is prejudicial to justice and finally to well-being, even for the poor, etc..

In order to facilitate the determination of the issues in certain complex situations, the higher political authorities may, to a certain extent, define in advance the terms of exchange of relevant values at differents levels. Value tables would be thus inserted in the computer programme so that the decision-maker at a lower rank can be helped by this way. But the risks of such a processing require precautions, even if the result of it is submitted to human appreciation. Anyhow, in sensitive decision fields, especially when human behaviour is concerned, this processing must be excluded as it will be further explained.


The assertion according to which there is an impossibility of modelizing axiology, efficiently and without risks and, consequently, of computerizing the making of the essential decisions concerning human beings and society, is not always fully accepted. Such a philosophical position is sometimes seen as a distrust towards the genius of our species, a refuse of technical progress and of the "social invention" and even the stamp of pusillanimous mind. But this assertion has not at all this negative bearing.

It has been exposed in the precedent part that, in spite of the "no-bridge", the arbitration or conciliations between values, although non modelizable, might be made by individual or collective decision makers. But such a possibility is limited or jeopardized by the absence of universal concepts so that the decision that are taken in such a situation are, to a certain extent fragile, insecure and subject to contestation. Nevertheless, this pessimistic view may be amended in consideration of new boundaries which can be sketched in the technological and organisational fields.


The axiologicaol decisions that can be made, so diligently prepared that might be, necessarily proceed from their environment. The cultures, beliefs, nations, political powers present an important diversity in their value systems which moreover are variable in the course of time. Blaise PASCAL said : "Vérité en deça des Pyrénées, erreur au delà" or "Plaisante justice qu'une rivière borne".

1. For, an adequate modelization of the axiological decision requires an exhaustive and operational knowledge of our world, of the human being and of the society where he lives. Much as such a modelization is thought as possible, those who affirm that refer to essentialistic conceptions and systems which can be also said as idealistic or transcendental. But these conceptions or systems, even we though cannot deny their rôle in our representations and comportments, finally appear to be unefficient, when not inhibiting as to the exigences about the acquisition of knowledge, the making of rational and equitable decisions and the efficiency of our undertakings.

As Gerald M. EDELMAN , Nobel prize of medicine, observes, the essentialistic conception that was advanced about the living species is unacceptable. The theory which has an explicative bearing is evolutionism, not fixism. More generally, we have to consider, without necessarily adhering to existential philosophy, that the existence precedes the essence, and not the contrary. Platonician philosophy, as well as the subsequent idealistic philosophies, whatever be their seduction, are too diverse, to uncertain, when not contradictory, to offer an universal "Weltanschauung " and a general model of individual and collective behaviour.

History as well as the current of events, in the end of this century, also show that nothing is to be expected from religions for that concern. Of course, it would be unbecoming to deny the importance of faith and even of religious fervor. But it remains that religions confronted each other, sometimes with violence, in the course of history. The theological debates, often of a high philosophical level, did not bring a large amount of clarity about the nature of man, his condition and his destiny. What is revealed to-day in the self-questioning of believers is anxiety or perplexity.

This impossibility of finding in religions a relevant model for our time is felt by those whose faith appears to be unalterable. André CHOURAQUI is learned in the Jewish, Christian an Moslem theologies which he comments with erudition and enthusiasm. In spite of such an oecumenical credo, he however writes : "In their oppositions, the religions appear to be dictatorial, authoritative, tyrannical, sometimes criminal, even in our times" . The Jewish philosopher, Hans JONAS questions himself about "the concept of God after Auschwitz . It brings out from his meditation that if God has not intervened in order to oppose to the "shoah", it is becaus that, since the creation of the world, he had renounced to be almighty, so that mankind be and remain free. In other terms, it is incumbent on ourselves and above all to the " just " to face evil and to be in search of good. As to the political doctrines, it is obvious they are not in a better situation in order to guide us, even though the political parties are indispensable in a democracy.

2. These limits, concerning philosophies, religions and political ideologies, should not be taken as a profession of "scientism" . The different sciences or science in general are no more apt to enlighten man about what he is, what is his hope in this world and what is his destiny . If sciences, as well as philosophies or religions, offer us some assistance, they are unable to dictate "categorical imperatives" . We especially have to throw away the so-called "anthropic principle" sometimes mentionned in cosmology, according to which the "universe" would have sprung up from "nothingness" and come off necessarily, after "billions of years", to man, the paramount being henceforth unchallenged.

Thus what is human, "too much human", as said Friedrich NIETZSCHE, cannot lead to a perfect intelligibility and therefore, escapes for an important part, to our capacity of modelization and automation.

3. If this attempt of demonstration does not appear to be fully convincing, it is possible to comfort it with more abstract arguments. Such an approaching suggests to put the question : Is a discourse about the "whole" (about the totality of things) relevant to science? That is the questionning of Etienne KLEIN .This query concerns not only physics, biology, human and social sciences, but also mind processes and, especially, the decision making submitted to a computer.

Before examining if a " theory of the whole" is possible and valable, we have to wonder wether this has a sense. For the "theory of the whole" requires that the human experts, or the computer system to create and programme, could have an image of "the whole", including themselves, which is unattainable. Morever, such a perspective is incompatible with the GÖDEL theorem of indecidability, according to which a theory cannot be founded entirely upon its own axioms. Far from building a "theory of the whole", the most remarkable minds did not easily triumph over the difficulties of demonstration that present certain theorems, the expression on which are very simple as "four colours suffice" or FERMAT'S conjecture.

4 .If we now consider a more modest and more practical level of knowledge, we observe that a number of new disciplines, which appeared in the second half of this century, did not succeed, in spite of appreciable results, to that was expected from them : for example, operations research, cybernetics, systems science, game theory, PPBS (planning, programming, budgeting system) , etc.. Finally it appears that as to the modelization and automation of the major decisions, we have to remain modest, without giving up reasonable ambitions.

Finally the limits also concern the computer and the conditions of its utilization. These limits are technical, praxeological and financial. In pratice, they often are connected or cumulative.

Up till now, it has been understood that the logical and arithmetical calculation did not raise important difficulties or obstacles. But this assumption become contestable when the matter is to computerize legal decisions that present a high degree of complexity. The difficulties concern, not only the hardware to be used, but the software also.

The praxeological limits principaly concern questions of time.The efficient modelization, especially in the field of economic and social prevision, requires time, which is often considererd as too long by the politician deciders. Even though a precise calculation is realizable, the required time is not compatible with the political exigences. Then, the experts must simplify their models which diminishes the quality of their work.

In addition, it is to be added than an accurate modelization can be very expensive. Its cost may be sometimes considered as excessive, if compared to the results of empiric valuations. It is often the case in legal affairs "rules of thumb" or "at arms length", methods are relatively cheap and efficient.


Man is "un être de projet" said Simone de BEAUVOIR. It is why it is not in our nature to abandon reasonable ambitions. So that the preceding considerations are not an expression of pessimism and, still less, of despair. The human beings and the societies they have constituted are in a position of deciding and acting, imperfectly, but usefully, what they have ever done, in attempting to progress, but generally remaining conscious of the relativity and uncertainty of their choices, if are excepted the fanatics, enslaved by the odious demon of certitude. "We are embarked" said Blaise PASCAL, scientist, philosopher and theologian. He thought that we have to decide and act without being in possession of an exhaustive knowledge, impossible to acquire and to master. It was, consequently, necessary that mankind should progressively constitute a corpus of rules, imperfect in absence of an ideal model, but sufficiently efficient and equitable. This corpus is a set of rules of law which emanate from the constitutional and legislative powers, the government and administrative authorities, the sense and bearing of these legal rules being on occasion elucidated by the courts.

1.The legal system is thus a model, the rules of which are assigned to the citizens, the entrepreneurs, the social organisms and to the political decision makers who have a major rôle in laws and regulations making. But this imperfect model cannot be made entirely explicit and has to evolute in order to be adapted in a changing society. It cannot be programmed and absorbed by a main frame computer so mighty it might be. The incompleteness of the legal system and of its computerial "translation" is principally due to the impossibility of determinating, in advance and with a sufficient accurateness, the terms of exchange between the socio-economic and ethico-political values.

But whatever be the importance of law in modern societies, it would be absurd to see in it a sort of "hypostasis". Law is only a human work and, in our days, essentially a lay, secular work, even though we observe religious remanences in certain countries. Law has made important progresses in Rome and if the roman legal thought is still a reference for us ; it is because law in Rome, has been soon laïcised : jus has been clearly distinguished from fas .

Law is not a science, in the strict acceptation of this word. According to the phrase proposed by Lucien SFEZ , law is a theorized practice . In other words, law is empirical, in the noble sense of this adjective : it proceeds from the confrontation with the reality (cf. Greek : empeiria, experience). It emerges from the things of life (les choses de la vie).

Law does not claim to transcendance or essentialism : it is existential. This quality makes its efficiency and its aptitude to univerzalisation. Law is now for us a new boundary. We observe to-day a bringing together between the law of the continental nations in Europe and the laws deriving from the English law, especially the law of the United States, which also influences the Japonese law and those of the South-Eastern Asiatic countries. Such an influence is sometimes negatively felt; nevertheless, it is wishable in many respects. We may notice a similar evolution as to human rights where the European legal thought played a major rôle. Finally it is not fortuitous if religions, in their integrist orientations, are obstacles, probably transitory, to the universalization of law. The progress of the legal decision, at an axiological level, will be favoured by the universalization of law.

2. This trend to a universalization of law perhaps is a promise as to a more audacious use of the computer in legal affairs.

Of course, it is understandable that the legislators in occidental countries have been prudent as to the computerization of legal decisions, and this prudence remains legitimous.

For what is a computer ? As Gerald EDELMAN explains, if the computer seems to be able to perform mental functions, this is only appearence or illusion. What is a computer ? It is a Turing machine, abstract model of an automate that is capable of processing algorithms. But these treatments consist in formal operations, without reference to signification of the data and results. Moreover, the more complex functions are completely outside of the capacities of the computer : to reason about that goes beyond the algorithms, to appreciate a situation that is not described in terms of formalized scheme and, overall, to invent, undertake, whish, hope, believe. As underlined by G. EDELMANN, the computer is unable to integrate intentional states into a mechanical treatment. Thought cannot be reduced to logic.

But new progresses can be made in the modelization and computerization of legal decisions. It is particulary important to envisage the cases where the legal decision to be taken require appreciations of human behaviours, expecially with reference to psychological profiles or schemes of personality. In such a situation, the decision maker has the duty of intervening deeply in the decision process. The aid of the computer is not prohibited. It can supply precise and objective data for the appreciation of behaviours. But the decision maker must himself, at least, proceeds to the final appreciation. If the decision is to be taken by a Court, the prohibition of a computerized process is stronger, especially in criminal trial (petty offences and minor infractions being excepted). Such is the situation in France, in pursuance of the law "Informatique et libertés". But even in absence of a written legal text, this prohibition seems to be imperative. Apparently, there is no example of the use of automation in the courts, about the matters that would concern felonies (crimes) or, more generally, major offences or in case of definition of psychological profiles or schemes of personnality. One may affirm that the analysis and appreciation of a human behaviour cannot result, for the essential, from a mere logical-deductive process .

This factual impossibility in practice, but which also is an ethical and legal interdict, does not find its reasons in a insufficient progress in technology and social sciences. It seems that it shall be a permanent situation, since it is very unlikely that time is to come in which, finally we exactly know what is man and what is a human society.

Nevertheless, there is perhaps as to the developments of computerization in court decisions a new boundary to conquer. If we consider the progresses in hardware and software, the interdict which concerns computerial processing of psychological profiles or shemes of personnality could be abolished, under the condition of a final examination by the judge or of the court. For the tribunals are overloaded. They often are obliged to deliberate with an excessive celerity, consecutive to the abundance of cases to be examined. A computerial preparation of the cases would be useful aid without damage for the defendant or accused. In this census of the new boundaries of the legal computerization, a reference must be made to connexionism. Significative progresses have been realized, in neuromimetic networks, especially in the legal fields. Of course, as Gerald EDELMANN explains , it is not a panacea, but there is in this new approach a promise of an important impact in legal automation.


For sake of brevity, but also in order to remain modest in a matter in rapid change, the so-called conclusion will only consist in thanks addressed to Mrs Patricia HASSET for her amiable invitation of which I was honoured and to Mrs Danièle BOURCIER. I also express my gratitude to the Syracuse University.

The writing of this lecture was for me an interesting opportunity of a deeper reflexion on a subject which often was for me a theme of meditation.

My thanks also are directed to the audience for its attention and patience.

Thank you cordially again to all of you.

© Réseau Européen Droit et Société