أبحاث ودراسات

A.I. Judge for the future? /Sara Nehme (قاضي الذكاء الإصطناعي كبديل في المستقبل!)

Sara Nehme:
Technology has brought major changes to our lives. It transformed humans from self-depending to machines-depending. This shift initiated lightly with simple automated robots and turned to techs that think and solve problems for humans. Technology and its continuous developments are no longer confined to one field; on the contrary it had entered and has been changing the environment and norms of all the fields out there, for example: FinTech, MedTech, EdTech, RealTech, and LegalTeh. Technology has already changed the practice of law, started with lawyers increasing the usage of A.I in the firm of predictive coding, predictive analysis and machine learning to the usage of AI systems for providing investigative assistance and automating decision-making processes in many judicial systems across the world. Three types of legal tech innovations have entered the practice so far: supportive technology innovation, replacement technology innovation and disruptive technology innovation. Questions rise on how efficient, helpful and reliable these technological innovations are? How much change in the judicial process and in the justice system can they bring? Are these changes problem bearing and if so can continuous development overcome them?
LegalTech wave reshaped the judicial system in 3 different ways for the time being. It introduced 3 types of tech innovations: supportive tech, replacement tech, and disruptive tech.
1. Supportive Tech:
It is invented to assist and support people who recourse to the judicial system and the people working in the judicial system. As for the 1st category, they obtain information online on rights, law, the judicial process, alternatives and locate justice services online. For instance: Online services| UK Department of Justice in can be accessed on jstice-ni.gov.uk.
As for the 2nd one, they use current documentary discovery programs that utilize predictive coding to read and analyze millions of pages of discovered documents and are able to select a relevant material in a matter of seconds.
2. Replacement Tech:
Technology’s continuous alteration in justice system includes replacing a physical court and litigation process with an online alternative that resolve dispute while retaining stature and power of a physical court. Even though online courts hit its epitome during corona virus and lockdown, but it didn’t abate afterwards but kept its pace if not increased worldwide.
For instance: In 2021 TIKE with the cooperation of Lebanese Ministry of Justice started a project, “equipping 12 Lebanese online courts” distributed all across Lebanese courthouses. This project took off in Baabda courthouse through zoom in a criminal case but on matters not more than bail and hiring attorney. The spectrum gets larger internationally where civil, criminal, administrative or other cases are tried online and from hearing sessions till the verdict come out.
3. Disruptive Tech:
In contrast to traditional courts where one verdict is the outcome, some sophisticated programs that focus on predictive analysis are designed to encourage the development of a number of options for one verdict. In this context, it is possible for A.I not only to support judges and arbitrators but to supplant them – in a lower case decisions making so far.
For instance: Alteras and his fellows’ program predicted the outcome of cases tried by the European Court of Human Rights based solely on textual content. The program (A.I judge) formulates a binary classification task where the input of its classifiers is the textual content extracted from a case and the target output is the convention of human rights. Textual information is representative using contiguous word sequence i.e N-grams and topics. This model can predict the court’s accuracy (79% on average).
Another example on ADR: Rechtwijzer is an advanced ADR program that incorporates ODR. It was 1st designed in 2007 to assist people on getting directions for a lawyer or judicial support then enhanced its services into dispute-solving. The current model cover consumer and relationship breakup in depth with “lite versions on employment, tenancy and administrative law issues”. Rechtwijzer asks the parties questions in multiple-choice form and provides options based on the input information. The program also provides information, tools, and links to other websites and personal advice which encourage the parties to resolve the dispute between them.
So legal tech is not just helpful and efficient but also needed when it comes to “supportive tech”. It can finish – the time-consuming and automated – jobs in a fraction of time that human labor would require. This leaves for the latter much time to finish important tasks they have on their hands. While the 2nd one “replacement tech” swings between a “step-forward for making judicial process easier” and “traditional way is more preferable due to the importance of eye-contact, voice shifting, and body-language role in trail”. However, the third one “disruptive tech” might impose certain problems that will be discussed in the following paragraphs.
Lawler predicted that one day the computer will be able to analyze and predict the outcome of the judicial decisions. More than 50 years later, the advances in natural language processing (NLP) and machine learning (ML) provided us with the tools to automatically analyze legal materials, so as to build successful predictive model of judicial incomes.
In order to understand the problems surrounding such programs (A. judge), one must understand how it works first. Essentially and in must programs and not all of them, the systems asks a number of questions or use existing data about users and cases and poses questions about the dispute to enable an accurate description of the dispute to be built, then forms a conclusion by applying the law to the dispute description. A.I does this by applying set of rules on specific set of facts. Finally, A.I can perform tasks based on description given. This process may enable indication decisions or even final decisions to be expressed. Such systems can be continuously updated and reflective, and machine learning enables system to improve and be constantly revised with the new data set.
Given a quick view over how A.I judge works with 2 examples before, we can now turn to discuss some major problems A.I might face:
1. Legitimacy:
In real daily life a judge is appointed to issue verdicts based on law, regulations and norms but in the case of the A.I judge, who possess the legal authority to issue such decisions? Is it the programmer, policy maker, or the automated system itself (A.I judge)?
2. Coding law and precedence:
Coding involves writing code (instructions) for the computer to interpret human language, process the instructions or commands and execute it. Yet coders are not policy or legal expert, so how can we make sure that complex regulations are currently transported into binary code? Programmers might make errors in this process, leading to automated decisions that diverge from the policy that A.I judge is meant to execute. Not to mention that if the legal rule was ambiguous and programmers took it upon themselves to make their own judgment call. Moreover, laws interpretation may hold different meanings depending on the context and sometimes which jurist’s opinion does the judge taking over the case follows. Furthermore, codes will need to be constantly updated due to frequent amendments, new case decisions, and complex transitional provisions.
And here is an example when A.I automated system caused a problem: In UK, post office’s accounting system, Horizon, controls the accounts of 11,500 post office branches around UK. There was a series of alleged frauds by sub-postmaster and sub-postmistresses. The latter took a civil action against post-office claiming that no fraud had taken place but rather discrepancies arose from system errors. They won the case in 2019, with judge criticizing the post office the post office and I.T supplier.
3. Discretionary or rigid decision?
Discretionary decisions may need to take into account values, the subject features of parties and any other surrounding circumstances that may be relevant; issues that some automated systems (A.I judges) are failing to take into account, since rules and regulations only are what their verdict is based on. This helps ensure consistency in decisions making instead of having divergence in precedence. Also, will prevent us from having a turn in precedence when new norms, point of view, and custom changes; though this dilemma is changing with the help of machine learning, deep learning and reinforced. Machine Learning focuses on data and algorithms to imitate the way that humans learn, gradually improving its accuracy.
However, programmers can build algorithms that have biased assumptions or limitations embedded in them or they can unconsciously phrase a question in a biased manner, which might lead to biases becoming embedded and reinforced through the process of machine learning. This poses a problem that the system being discriminatory; a problem that courts worldwide already witnesses with human judges.
All in all, Technology’s development is faster, greater, quicker and more precise than ever. And even though A.I was invented to aid and support us, its continuous development enabled it to take over some humans roles. This enabled role is scoring important and efficient progresses and results in many fields like MedTech, Ed Tech, AgriTech. Will these results be seen in a more complex and empathetic field as the legal field? As machine learning and reinforced learning are evolving with time, they are making what seems impossible possible. And as Chief Justice John G Roberts Jr (USA) when asked “can you foresee a day, when smart machines, driven with artificial intelligences, will assist with courtroom fact-finding or, more controversially even, judicial decision-making?, responded “it’s a day that’s here, and it’s putting a significant strain on how the judiciary goes about doing things”.

Daniel Kirsch and Judith Hurwitz, Machine Learning, published by John Wiley & Sons, inc, 111 River St. Hoboken, 2018
Tania Sourdin, Judges, Technology and Artificial Intelligence, published by Edward Elgar Publishing limited, 2021
Melissa Perry, iDecide: Digital Pathway to Decision, Federal court of Australia, published on 23 March, available at shorturl.at/bfwJ1
Tania Sourdin, Judge v Robot? Artificial Intelligence and Judicial Decision making , Handbook for Judicial Office, available on shorturl.at/blQ07
Anwesha Barari, What is coding? , EMERITUS, published 15 July 2022, available on shorturl.at/lrEJP
James Christie, The post office Horizon IT Scandal, James Christie’s Blog, available at shorturl.at/jo568
HiiL & the Dutch Legal Board , The Rechtwijzer Justice , Justice innovation Stanford Legal Design Lab, available at shorturl.at/prW68

“محكمة” – السبت في 2023/1/14

مقالات ذات صلة

اترك تعليقاً

لن يتم نشر عنوان بريدك الإلكتروني.

زر الذهاب إلى الأعلى
error: Content is protected !!