A few years ago, we were excited to see that increasingly more aspects of our lives started to become computerized. From the way we buy our bus tickets to the way we interact with one another, more and more daily activities shifted from their traditional approaches to some that are, one way or another, dependent on modern technology.
It was thus obvious that people started to wonder whether or not everything can eventually be computerized. More importantly, will we ever get to a point where judicial systems, which are naturally based on human activity, are going to be computerized? Would it be actually better this way?
We invite you to read the following debate and decide for yourselves which side is the most persuasive and convincing. We hope that you will find it just as interesting as we did!
Cristian Ioan
Cristian is a 4th year law student at Babeș-Bolyai University in Cluj-Napoca, Romania. His biggest desire is to deepen his knowledge of law and bring his contribution to help others understand and respect it.
Dorin Giurgi
Dorin is a 4th year law student at Babeș-Bolyai University in Cluj-Napoca, Romania. He aims to become a lawyer and to contribute to the predictability of the Romanian judicial system in his areas of practice.
Opening Statement - Cristian
For the purpose of this debate, I will not dive in the technological aspects concerning the current capabilities of computer systems. I will rather consider a theoretical model in which computers are able to deliver case law at a similar judicial error rate as the current one. Whether that is technologically achievable remains debatable, and only time can tell.
Firstly, depersonalizing the judicial process would greatly benefit the system as far as correcting problems concerning corruption related issues. I believe that there is no significant reason for further elaborating this point, as corruption is fundamentally humane and can even be regarded as a cultural issue. Even though other measures can hinder this phenomenon, switching to a computer- based judiciary procedure would mostly eradicate corruption in this field in the swiftest and most cost-effective manner possible. Switching to a more utilitarian model will also solve a plethora of issues that derive from the human nature of those involved in the judicial process. It is not hard to see how race bias or bias based on looks, religion or sexual orientation would be eradicated.
Secondly, the judicial process would be significantly less time consuming, as delivering a decision can be a matter of minutes. Basing our theoretical model in a civil law legal environment, and taking the Romanian system as an example, in contract law related cases, where hearings are not fundamental and the parties usually do not personally attend them, once the facts have been established, a decision is easily delivered. Moreover, the whole process of going through an appeal would be rendered superfluous. Thus, Appellate Courts would mainly be concerned with assessing whether the case facts have changed, based on new evidence.
Furthermore, the overall cost of going through the judicial process would greatly decrease. Thus, the divide between individuals who can or cannot afford costly legal services would be closed. Assuming that the main goal of the judiciary is to achieve a fair judgement, and that real, and not only procedural, equality of arms is essential to achieve such judgement, this is an adequate change in that direction. Currently, even if a person has the right to legal assistance (although, usually only in criminal matters), even if the assigned lawyer is highly competent (which mean having a similar legal prowess to a top attorney), it is almost impossible for him to achieve the same level as a top tier law firm working on a case. Setting aside the fact that quality lawyers tend to have high financial demands, which make them inaccessible to a regular person, simply having the possibility of dividing the workload makes a significant change.
Finally, removing personal opinions and empathy from the matter should result in less cases where people knowingly act contrary to legal provisions, on the hope that their case will be different. Although judges should rely on the spirit of the law when analysing the facts of a case, such recourse, if it contradicts or exceeds the letter of the law, should be cautious. It is also based on years of experience and study. Yet, discouraging other subjects of law from acting outside the letter of the law should be seen as essential to a better society. The coldness of a machine seems to be incompatible with the belief that their actions could be seen from a different angle than the one that the law declares.
Opening Statement - Dorin
First of all, we acknowledge that in order to fully grasp the subtleties of this topic, a definition of terms is necessary. Therefore, by computerised, I understand all kinds of electronic systems that are capable of applying certain rules (the law) to real-life situations. Considering the implied premise of the debate, that of the existence of such computerised system, I will only mention that, for the purposes of this debate, this system does not have the emotional and social processes of a human being. If someone would hold the contrary, and we would further accept as a premise that the system possesses perfect understanding of human emotions and behaviour, the opposition would essentially have to argue against an entity that is like a human, only that it is a great number of times more efficient.
My position in this debate is that a) fully replacing lawyers is not possible, and, even if such system existed, b) neither that (replacement) of judges.
Regarding a), the difficulties one finds when trying to replace lawyers, the following objections can be raised:
i) Non-jurists will find difficulties in accessing the "computerised system";
ii) The need of assistance regarding procedural requirements, e.g. as to what constitutes evidence, may not be fullfilled;
iii) The need for a legal opinion as to possible outcomes and courses of action, is best satisfied by a human being (because you can relate to other persons).
It might be true that, if such system existed, court lawyers will become less relevant or inexistent, since the system will "think" automatically, without the need for contradictory pleadings. To summarise, I hold the belief that lawyers will be the link between non-jurists and the "judge", translating the facts of the different cases in legal language, for the judgement to take place according to the established algorithms.
Now, I will look at the second part of the debate, "would the judicial system be better if judges were replaced by computerised systems".
Regarding b), I believe the following arguments will find relevance: i) Laws cannot be applied without interpretation, ii) Lack of second instance examination, iii) Limited social and emotional intelligence and iv) Cybersecurity risks.
i) At this point, I would like to propose the following ideas: Language is not logically perfect, therefore, laws will not be perfect, meaning that interpretation will be required; Interpretation is a very complex endeavour, meaning that different criteria and sources will need to be observed, some of them involving a subjective appreciation, e.g. the ratione legis approach; supposing that a computerised system would be able to understand such complex realities (i.e. which need to be approached from multiple perspectives) would be unrealistic, since, in such a situation we would suppose the existence of a "supreme being", which can think on its own using a complete understanding of the laws of nature.
ii) It is not clear how errors of this system would be addressed, and whether a "second instance" would exist. If the latter is applicable, the question that can be raised is how this "second instance" is different than the first one.
iii) Law is not applied in abstract, but in relation to facts (including witnesses, testimonies etc.). It is hard to believe how justice will be made without the judge screening the person behind a testimony.
iv) A computerised judicial system would be subject to a great number of cyberattacks. Considering that even the most secure systems have been repeatedly breached, I raise the issue of this happening in the context of a computerised judicial system, with serious consequences.
Cristian argues that a computerised judicial system is one which is not corrupt, corruption being a problem inherent in a human based system. Also, endlessly long trials and costly judicial services will be a thing of past, because computers are both time and money efficient. Lastly, the risk of personal opinions and beliefs affecting the outcome of a trial won't exist, because the coldness of a machine will make it rely on the letter of law. However, Dorin argues that such a system won't be able to respond accordingly to people's needs, such as the need of assistance regarding procedural requirements, or the need for a legal opinion as to possible outcomes and courses of action. Moreover, a judge doesn't only read the law, he interprets it, and if he makes a mistake, there is always the possibility of accessing a second instance. Machines cannot be capable of such things, and their lack of emotional and social intelligence along with the risks of cyberattacks are strong motives against implementing this kind of system.
Rebuttal - Cristian
It has been argued that non- jurist will face issues when accessing the computerised system. There are several arguments to refute this statement. Firstly, this problem can essentially be solved through a user- friendly interface, as it is more of an issue concerning luddites, rather than non- jurists. Secondly, the term of comparison we should refer to when assessing the difficulty of using an AI system must be the legal system that is currently in place. Thus, the question should be whether non- jurists would face issues that are more considerable when using computerised system compared to the current one. Since the system is structured on an automatic process, the answer is negative. Moreover, if a person cannot use the system, as a last resort, he or she can recourse to specialist advice. The system would still be more affordable, efficient and faster than the current one.
It has also been stated that there would be problems concerning the need of assistance regarding procedural requirements. Taking into consideration the fact that procedural requirements are usually positivistic legal notions, the system should have the capacity of informing the person on these requirements and automatically applying them to the specifics of a case. Since the system crunches data and compares it to legal provision, it is facile to council the person on possible outcomes. Thus, a person can make an educated decision, whether he should invest his time and money in a trial. Since the computerised lawyers will work on algorithms, thus being objective when setting expectations, I believe that would be an improvement on the current system, not a problem.
Regarding the aspects concerning the replacement of judges there were four main arguments. Firstly, the fact that interpretation is a key component of applying the law does not mean that it cannot be quantified when creating an algorithm. In fact, where language fails, a majority- accepted interpretation could be integrated in the system. This would also serve the principle of legal certainty and legitimate expectations. Having a unified interpretation is a sought after feature of a legal system. The role of legal doctrine will actually increase. Ratione legis is just a manner of teleological interpretation, which can be programmed in an algorithm, as long as it is specifically applied to a certain norm. The AI should not interpret the law itself, based on principles as broad as ratione legis interpretation. Rather it should apply the law and accepted interpretations to the specifics of a case.
While a second-degree court would become superfluous in a vast majority of cases, a second verification of the logical accuracy of the applied algorithms should be instituted. Furthermore, other evidence can be examined, and many issues that are addressed through extraordinary recourses can be addressed through an ordinary appeal. Even if legally trained human judges would scrutinize the decisions, their task would be considerably simplified, and the duration of the trial would be abridged.
The premise that limited social and emotional intelligence would be an issue is quite debatable. It means that the system would be more objective. It is highly debatable if judges should let their subjective emotional appreciation eclipse actual legal provisions. For example, other evidence should contradict testimonies in order to be rendered nugatory. The judge's biased opinion on a person should not be seen as a legal tool.
Concerning cybersecurity, such an issue would still be significantly less poignant that corruption and work- related criminality. Moreover, any person that tries to hack the system would still be subjected to justice. Finally, the only real risk would be an attack from a nation state actor, and those would be grave perpetrations of international law, that states will avoid, considering the consequences.
Rebuttal - Dorin
It has been said that corruption would disappear. My first response is that (i) replacing judges and court lawyers does not mean that the entire judicial system will function without humans. We will still need at least cops to do the field work (e.g. gathering evidence), lawyers or jurists to enter the data into the system, and prosecutors. The phenomenon of corruption would still be possible at these stages, since it has not been proven yet that they will disappear.
My second objection is that (ii) the effects of eradicating corruption would be nullified by problems regarding cybersecurity. In short, cyberattacks, which I believe to be inevitable, have the potential of creating more damage than corruption, as claimed to exist in the status quo.
Furthermore, it has then been said that other issues would be solved, like: race bias, bias based on religion, sexual orientation, looks etc. In my opinion, on one hand, (i) a human judge would be accused to have a "race bias" if he treats differently persons of different skin colours. On the other hand, (ii) the computerised system would use statistics, which would indicate that one group is more prone to violent crimes than the other. In this case, what is the computer going to pick?
Moreover, if the computerised system would be an artificial intelligence, how can we make sure that the computer judge will not develop other criteria for solving the cases, other than those provided by its creator? For a human judge, most of the times, due to a certain length of the judicial process, biases are uncovered through the judge's behaviour. What can you do in case of an AI?
With regards to speeding the judicial process, I would remind that in the present system, once the facts have been established, a ruling is going to be pronounced at the end of the judicial debates. In my opinion, the existence of cases in which the reasoning is uncovered only a few months after the ruling is not a valid argument against the current, but rather an indication of the low performance of the observed legal system (i.e. Romanian) regarding the work load of the judges.
Additionally, I would like to direct the readers to the arguments I presented supra regarding hearings and appreciation of facts, and argue that these are processes which are going to be very difficult in a computerised system.
With regard to the appeals procedure, it is hard to imagine that computers will not make mistakes, especially when we consider the complexity of the matters subject to analysis by the computer judges. As I pointed supra, it is unclear what remedies would be available for a party who identifies mistakes in a ruling. Moreover, it is unclear which entity would judge such an appeal, and why its judgement would be better than the previous one.
Moreover, the statement overall cost would greatly decrease is based on the assumption that maintaining a computerised judicial system functional is not going to require costs. I beg to differ, as the system would need databases, improvement of its security, technicians, updates and so on.
Finally, as explained above, the coldness of a machine is incompatible with imperfect legal norms, which are inevitable. As it has been shown, language is not logically perfect, therefore, a legal system based on language is not going to be logically perfect either, and would require interpretation. On the other hand, how would a computer apply contradictory provisions, i.e. interpret the law? In these situations, is the purpose of the law ever going to be achieved?
At this point, Cristian argues that problems such as the risk of non-jurists having issues accessing the system are not problems which are non-solvable or bigger than the ones the judicial system already has. He also states that a computerised system is based on algorithms, so explaining possible outcomes of a case will come very easily, because this is what those algorithms do: they crunch and compare data. Moreover, he consolidates his previous ideas, by pointing out that the lack of personal interpretation will lead to a more predictable and objective legal system, and if the only problem this system would have is the risk of cyberattacks, this is still less harmful and more inclined to be solved than corruption.
On the other hand, Dorin points out that humans will still be needed in the system, meaning that corruption cannot fully disappear. He also raises the problem of computers' capabilities and our lack of methods of tracking down their possible biases, further indicating that current problems such as extremely long trials are not necessarily related to human judges. Finally, he restates his main arguments, revealing that the lack of an efficient appeal procedure in a computerised legal system (which is essentially built on something interpretable-language) could lead to mistakes that can remain undone.
I hope this debate was a pleasure to read, and the conclusions will help you form your own opinion!
Conclusions - Cristian
Bearing in mind all the legal arguments that were raised, I hold my belief that replacing both lawyers and judges with computerised systems would greatly improve the manner in which justice works. As I have previously mentioned, the system would be more equitable. Even when taking into consideration Dorin's correct point that the cost of a functioning computerized judiciary would be hard to estimate, I find that the system would still be more accessible, as the cost of lawyering would decrease. Moreover, I find that having objective assessments concerning how the law is applied should be perceived as a desiderate, rather than an issue.
Tracking back to the question on which the debate is constructed, we should bear in mind that it focuses only on judges and lawyers. Replacing them does not mean the justice system will no longer need humans. Rather, it means that the role of human implication in the process will change. This does not mean computers will determine the law. It only means that they will apply it, according to the rules that humans set, according to their social needs. Supervision of the system will be necessary.
Indeed, changing the system we used for thousands of years is a big step. However, we should not fear change, as it is the motor of improving our society.
Conclusions - Dorin
Firstly, when talking about the (i) replacement of lawyers, I find it unreasonable to imagine such a software that would understand non-legal terms and explanations, which are sometimes incomplete and incoherent, depending on the level of education of the person submitting the complaint.
Secondly, when it comes to (ii) interpretation, I believe that once we accept that interpretation should be included in the legal norms, we would accept the idea of a case-by-case constructed legislation, trying to cover all the possible "misinterpretations".
Thirdly, regarding (iii) appeals, I argue that human judges would have the task of verifying the whole sentence, including the reasoning of the decision, basically having a similar role to the one they have today.
Fourthly, concerning (iv) cyberattacks, I underline that these have the potential of causing greater damages through less effort than corruption.
Furthermore, the (v) corruption issue, as I have shown, is not clear how will disappear, since humans will still be a great part of the judicial system.
Moreover, in the matter of (vi) biases, I argued that bias in humans can be discovered and dismissed in time, while bias in computers (see above) can't.
Finally, in the matter of (vii) social and emotional intelligence, it cannot be said that the system would be more objective, but rather that it would not function properly, e.g. humans will pick non-verbal cues indicating that one witness is lying, while computers won't.
Disclaimer: The arguments presented in this debate do not represent the opinions of the two authors.
This debate has been published in Lawyr.it Vol. 5 Ed. 2. All references used can be found at the end of that issue.
