«ПЕРСПЕКТИВЫ РАЗВИТИЯ МЕЖДУНАРОДНОГО КОММЕРЧЕСКОГО
АРБИТРАЖА В УЗБЕКИСТАНЕ»
Сборник международной научно-практической конференции
115
Sardor Bozarov
PhD in Law, Acting Professor, Private International Law Department,
Tashkent State University of Law
https://doi.org/10.47689/978-9943-7818-6-3/iss1-pp115-119
INTERNATIONAL ARBITRATION, HOW ARTIFICIAL INTELLIGENCE
WILL CHANGE DISPUTE RESOLUTION
Abstract.
The author considers the concept of artificial intelligence in this
article. In addition, the issues of implementing of artificial intelligence in
international arbitration have been analyzed. Besides, the author considers the
aspects of using of big data in international arbitration. In conclusion, the author
compares the advantages as well as the negative features of artificial arbitrators.
Key words:
artificial intelligence, machine learning, international
arbitration, prediction, advice, robotics, artificial arbitrator.
Сардор Бозаров
К.ю.н., и.о. профессора, Кафедра «Международное частное право»,
Ташкентский государственный юридический университет
МЕЖДУНАРОДНЫЙ АРБИТРАЖ: КАК ИСКУССТВЕННЫЙ
ИНТЕЛЛЕКТ ИЗМЕНИТ РАЗРЕШЕНИЕ СПОРОВ
Аннотация.
В этой статье автор подробно изучает концепцию
искусственного интеллекта и его применение в сфере международного
арбитража. Особое внимание уделено анализу использования больших
данных в арбитражном процессе. В заключительной части статьи
проводится сравнительный анализ преимуществ и возможных
недостатков применения искусственных арбитров.
Ключевые слова:
искусственный интеллект, машинное обучение,
международный
арбитраж,
прогнозирование,
консультирование,
робототехника, искусственный арбитр.
Artificial intelligence is the most important element of the Fourth Industrial
Revolution that has begun. Currently, artificial intelligence (AI) is considered one
of the most important areas of IT research, the driver of breakthrough industrial
growth, which has recently been called “Industry 4.0”.
Artificial intelligence has a relatively long history, dating back to Turing’s
theoretical research on cybernetics dating back to the early 20th century. And the
conceptual premises appeared even earlier – from the philosophical work René
«ПЕРСПЕКТИВЫ РАЗВИТИЯ МЕЖДУНАРОДНОГО КОММЕРЧЕСКОГО
АРБИТРАЖА В УЗБЕКИСТАНЕ»
Сборник международной научно-практической конференции
116
Descartes’ Discourse on Method (1637) and Thomas Hobbes’ Human Nature (1640).
In the 1830s, English mathematician Charles Babbage proposed the idea of
a complex digital calculator, an analytical machine that the developer claimed
could calculate moves for playing chess. In 1914 the director of one of the Spanish
technical institutes, Leonardo Torres Quevedo, made an electromechanical device
capable of playing the simplest chess endgames almost as well as a person.
In the 1830s, English mathematician Charles Babbage proposed the idea of
a complex digital calculator, an analytical machine that could calculate moves for
playing chess. In 1914, the director of one of the Spanish technical institutes,
Leonardo Torres Quevedo, made an electromechanical device capable of playing
the simplest chess endgames almost as well as a person.
Before comprehensively analyzing the peculiarities of AI it is necessary to
determine what is AI. AI has many definitions at present. The Oxford Dictionary
defines AI as “the history and development of computer systems capable of
performing tasks that normally require human intelligence, such as visual
perception, speech recognition, decision making, and translation from one
language to another” [1. P. 1].
Scientist John McCarthy – may have coined the term “Artificial intelligence”
– describes it as the ability to “make a machine behave in a way that would be
called reasonable if a person behaved that way” [2. P. 1]. Interestingly, both
definitions use human intelligence as a guide. They describe AI in comparison to
“tasks that normally require human intelligence” and “in ways that would be
called intelligent if a human behaved that way”.
Artificial intelligence (AI) refers to the simulation of human intelligence in
machines that are programmed to think like humans and mimic their actions.
The term may also be applied to any machine that exhibits traits associated with
a human mind such as learning and problem-solving [3].
Today artificial intelligence is implemented in the whole sphere of our life.
The use of AI in international arbitration has its peculiarities. If we consider the
issue from various points of view as a lawyer, as an arbitrator as well as the
legislator we can see the prospects of using artificial intelligence in legislature and
international arbitration. McKinsey developed an online test with the question
“Can a robot take my job?” This online allows anyone to indicate their professions
and count the possibility of being replaced by a robot [4. P. 1].
According to the result “23% of your lawyer’s work can be done by a robot”
and this profession is “safer than 67.9% of other professions”. Tasks of lawyers,
which, as the test suggests, are within the power of robots include “the study of
relevant legal materials” and “preparation of legal documents”.
AI research tools based on big data combined with powerful search engines
can search for the “definition of an arbitration agreement under the laws of a
particular country” or “the latest Swiss jurisprudence regarding the waiver of the
right to set aside arbitral awards”, and the machine will find the relevant
jurisprudence, legislative provisions, and scientific articles. Some of these
programs include deep learning tools. They will continually improve their
«ПЕРСПЕКТИВЫ РАЗВИТИЯ МЕЖДУНАРОДНОГО КОММЕРЧЕСКОГО
АРБИТРАЖА В УЗБЕКИСТАНЕ»
Сборник международной научно-практической конференции
117
performance by asking for feedback after each research assignment and learning
from their own mistakes.
Some developers claim that their systems give a satisfactory answer 97% of
the time. Such tools can reduce research time, but human input is still needed and
should not be underestimated. There must be a live operator who knows how to
use the research tool, who can ask the right question, and who can analyze and
interpret the results.
Such tools can reduce research time; however, human input is still needed
and should not be underestimated. There must be a live operator who knows how
to use the research tool, who can ask the right question, and who can analyze and
interpret the results.
Today, big data processing has become an important aspect of any dispute
resolution mechanism, including arbitration.
This refers to the use of special programs to organize, analyze and process
large datasets that traditional databases cannot “Вigest”. They use predictive coding
to reduce the number of irrelevant documents that need to be reviewed manually.
Such programs are used, for example, in processes of electronic data discovery.
Moreover, the use of AI is also important in the field of predicting results. In
other words, it is very useful for large businesses because it can determine the
most likely outcome of future litigation” Of course, the question arises whether
computer programs able to predict the outcome of disputes. Recent studies have
shown that yes.
For example, in 2016, the research team collected all cases heard by the
European Court of Human Rights under Art. 3, 6, and 8 of the European
Convention on Human Rights, developed a data set, and offered a computer
program to analyze this array.
The AI had to keep track of the frequency, sequence, and clusters of words,
and then assign them importance according to compared with the result of the
examination of the case, whether or not a violation of the relevant provision of the
ECHR was eventually found.
The program was looking for relationships between words, their sequences,
and clusters that could predict the outcome of a case.
The research team then applied the program to other cases that had not
previously been entered into the system (that is, those for which the AI did not know
the result). The final accuracy of forecasts was 79%. The second study was conducted
in the 2017 year and concerned the decisions of the US Supreme Court [5. P. 2].
As in the first case, the training program gave deeds of certain years and
then asked to predict results for other years. The success rate was somewhat
lower, with the AI guessing 70% of the time.
Although the accuracy of the forecasts in this study is less than in the
previous one, it is more impressive: here the analysis touched on all branches of
law on which the US Supreme Court decides – a much broader scope than specific
legal issues Art. 3, 6 and 8 of the European Convention on Human Rights.
Commercial case prediction software is already being used in many areas,
«ПЕРСПЕКТИВЫ РАЗВИТИЯ МЕЖДУНАРОДНОГО КОММЕРЧЕСКОГО
АРБИТРАЖА В УЗБЕКИСТАНЕ»
Сборник международной научно-практической конференции
118
including intellectual property disputes, labor law disputes, etc.
However, there is not yet the technology that would be used in international
arbitration. One obvious barrier is confidentiality: access to international
arbitration decisions is much more difficult than access to decisions of the
European Court of Human Rights or the US Supreme Court.
This makes it difficult for the program to learn how to predict outcomes effectively.
This restriction applies to international commercial arbitration but to a lesser extent to
investment arbitration, where decisions are published more frequently.
Examining the ability of programs to predict accurately the outcome of
cases leads to the more fundamental question of whether decision-makers as
judges or arbitrators be effectively replaced by AI. Beyond the barrier of privacy
of court decisions (and thus reduced datasets) that AI have already mentioned,
there are other difficulties.
International arbitration cases tend to be complex, full of facts, and subject
to various applicable laws. Due to the non-recurring nature of international
arbitration proceedings, AI remain skeptical about replacing international
arbitrators with artificial intelligence any time soon.
There are other, perhaps more complex issues. AI will eventually be able to
resolve international arbitration cases. Will it be good? Will programs become
better decision-makers than humans? Here, several arguments can be made in
favor of the superiority of AI over humans.
AI does not experience hunger and does not have any feelings or emotions,
which means that it is not influenced by irrational factors when making decisions.
A group of scholars looked at cases of parole in criminal law and tried to analyze
what influenced the judge’s decision. Of the many factors, one turned out to be purely
irrational: the issue was about the decision, made by the judge before or after the
lunch break. Before dinner, the judge was hungry and more likely to refuse the
petition, but after dinner, he was more likely to satisfy him. However, there are many
arguments against using artificial intelligence to make decisions. The first possible
counter-argument in researching this topic was this: Judge Programs are dangerous
because they can give too much power to programmers. AI will make decisions based
on initial algorithms, which are determined by the programmers. In other words, the
one who prescribes the algorithm determines the outcome of the case.
Notwithstanding, Strong AI actually means, that the program learns itself.
In research on the decisions of the European Court of Human Rights and the US
Supreme Court, programmers did not set the algorithm, rather, the program itself
described the conclusions it would draw from certain findings in decisions. No
programmer has specified this beforehand.
However, there follows a second, perhaps more complex, argument against
artificial arbiters. No one knows why or how AI arrives to make precise and right
decisions. What conclusions does the program make in order to predict the decision
of the European Court of Human Rights with 80% accuracy? The AI bases the
prediction on the analysis of previous decisions, the appearance of certain words, or a
collection of words, but no one can explain in detail exactly how this happens.
«ПЕРСПЕКТИВЫ РАЗВИТИЯ МЕЖДУНАРОДНОГО КОММЕРЧЕСКОГО
АРБИТРАЖА В УЗБЕКИСТАНЕ»
Сборник международной научно-практической конференции
119
A group of sophisticated American academics has developed a machine
learning application, which claims to be able to predict the outcome of a case at
the Supreme Court of the United States (SCOTUS) with an accuracy of 70.2%, and
the voting behavior of individual judges with 71.9% accuracy [6. P. 5].
The prediction in criminal cases is another case study of the use of AI in the
United States. In order to predict recidivism in criminal cases, another group of
American scientists created tools such as the Correctional Offender Profiling for
Alternative Sanctions (COMPAS). Criminal judges in some states use this tool
when they assess the risk of recidivism of defendants or convicted persons when
making decisions on pre-trial detention, sentencing, or early release. The COMPAS
helps to reduce the number of detainees because these tools assess recidivism
risk more objectively [7. P. 5].
These examples clearly show that AI may be able to make accurate
predictions, but it will not be able to explain (at least to humans) how and why it
achieved certain successful outcomes. However, justification is one of the
fundamental characteristics of the decision-making process in both national
courts and arbitration tribunals, as well as justification, is the main principle of
ensuring justice. In addition, reasons on the cases allow the losing side to
understand why lost, and thereby make the decision more acceptable.
Reasons help the parties to change their behavior in the future. Finally,
a reasoned decision, if published, gives other tribunals the opportunity to follow
the same rationale or explain their deviation from the previous precedent.
By analyzing the whole abovementioned, it should be noted that artificial
arbitrators will not be able to justify the reasons for their decisions, which means
they will not fulfill the fundamental requirements of justice. In addition, of course,
this tendency leads us to make the conclusion that artificial arbitrators cannot
replace human arbitrators.
References
1.
https://en.oxforddictionaries.com/definition/artificial_intelligence.
2.
Каплан Д. Искусственный интеллект: что нужно знать всем (2016).
3.
https://www.investopedia.com/terms/a/artificial-intelligence-ai.asp
4.
Джонсон Д. Узнайте, отберет ли робот вашу работу. URL:
http://time.com/4742543/robots-jobs-machines-work.
5.
Aletras N, Tsarapatsanis D, Preoţiuc-Pietro D, Lampos V. 2016.
‘Predicting judicial decisions of the European Court of Human Rights: a Natural
Language Processing perspective’, PeerJ Computer Science 2:e93
https://doi.org/10.7717/peerj-cs.93.
6.
A. D. (Dory) Reiling, ‘Courts and Artificial Intelligence (2020) 11(2)
International Journal for Court
7.
Institute for Crime and Justice Policy Research World Prison Brief,
prisonstudies.org last visited on 6 December 2019.