Artificial intelligence (“AI”) refers to technologically created intelligence that is able to learn and process enormous amounts of data by using certain algorithms that compute much faster and with greater aptitude than human brains can manage.[1] With the latest improvements in auto-learning and the self-improving properties of artificial intelligence, AIs have started to replace humans in several industries, thus reducing the “human error” in many of the procedures they are involved. AI has become such a useful facilitator around the world today that even conservative areas like the legal field are in the process of implementing it to the extent possible. However, this gives rise to a number of questions. How would it be possible to implement AI in arbitration? Is AI likely to replace lawyers and/or arbitrators in the near future, or is being applied as a tool in arbitration the furthest AI can get? In this article, those questions will be examined briefly along with examples of current uses of AI in arbitration as well as the Turkish law perspective concerning the enforcement of arbitral awards rendered by AI.
The Use of AI in Arbitration: AI Tools
Artificial intelligence, as one of the main drivers of emerging technologies, has already begun to ease our lives including in our professions. Lawyers, arbitrators, practitioners, and clients around the world are now aware of AI more than ever. Although AI-powered arbitration exists on a global perspective, to our knowledge, it has not yet reached the extent that arbitration procedures can be solely conducted by AI without any human involvement. Today, AI is merely used as a tool to assist lawyers in arbitral procedures. Most commonly, AIs are now used to speed up paperwork and data analysis, but they may also handle different tasks based on each specific case. Some do research and summarize legislation while others provide statistics based on their examination of precedents on the subject and analyze whatever input the parties provide in the case.[2] With each use of AI, the hours required for lawyers to complete paperwork and case review are substantially reduced. Likewise, the AI also helps clients by decreasing lawyers’ billable hours and reducing the cost of legal consultancy services. Some of the most famous examples of AI tools include Lex Machina, Arbilex, Arbitrator Research Tool (ART), Liti-gate, and Arbitrator Intelligence.
All of these tools fulfill the duty of analyzing vast amounts of digital data to reach certain statistics and conclusions based on that data and help the case at hand.[3] With ready-to-use and reliable statistics and concrete data from former cases presented by the AI, lawyers can build their case on a solid basis and the possibility of success increases in high percentages. Overall, AI tools provide data-mining facilities which help ease preparations for arbitral proceedings.
One Step Further: AI Arbitrators?
Apart from AI tools, some AIs can also undertake certain duties for arbitrators where it is permitted in the legal system. Many questions and contradictions regarding the use of AIs in arbitration derive from the idea that an AI can act as an arbitrator and render awards. A non-human machine-based creation determining the destinies of legal cases for humans has caused hesitation from both a technical and moral perspective. There are many discussions and debates over AI acting as a judge and/or an arbitrator, such as the AI’s responsibility in cases of breach of confidentiality and liability for its actions and ethical concerns. There is also a dilemma over who should be held responsible when an AI-generated error causes harm to others. Since AI does not have the capacity to have rights or the capacity to act, therefore lacking the criminal capacity for its wrongful acts, this may give rise to convoluted situations.[4] Hence, any leak of information, faulty evaluation, or damages caused by an AI leads to a dead end when it comes to claiming damages for such defects or negligence. Yet, for the purposes of this article, we will set aside the theoretical dilemmas and solely discuss the current and concrete uses of AI arbitrators and the readiness of the legal systems for such a proactive phenomenon. The idea of an AI arbitrator can be broken down in three different possibilities:[5]
- AI completely replaces human arbitrators, does all of the work itself, and directly renders the award without any human effect or control mechanism;
- AI arbitrators and human arbitrators work in a combined tribunal and the final decision is determined by a common decree that they come up with jointly. This way the procedure is not completely out of human reach, but also remains as mechanic and computerized as possible. Therefore, any possible non-humane result can be prevented by the human arbitrators whereas possible human errors can be detected by the AI arbitrators; and
- Human arbitrators are the ones actually in charge of rendering an award, but there are also AI arbitrators present in the role of an advisory resort so that the human arbitrators can check their decisions.[6]
In today’s practice, it mainly seems to be operating in accordance with the third version of the AI arbitrator as described above so the role is limited to the AI assisting human judges, similar to the case of lawyers, reviewed above. For example, the Colombian government implemented an AI called “Siarelis” in its legal system to resolve corporate disputes. Siarelis acts as a guide for judges by questioning the case and providing a legal opinion on the case depending on the merits.[7] Likewise, Spain uses Jurimetria, “a legal prediction tool that works upon the analysis of millions of court decisions.”[8] Needless to say, judges are free to disregard the AI’s opinions and analysis.
AIs require a substantial level of progress to be sufficiently filled with the adequate experience and knowledge to reach a humane level of understanding with respect to a “sense of justice” and will create their own legal notion on how to solve disputes. Apart from how the abilities of AIs can develop as swift as the wind along with today’s technology, implementing the notion of justice soars beyond the capacity of algorithms since human virtues such as justice are too far-fetched to possess, ironically, even for humans.[9] Only when AIs manage to possess certain human virtues will they be more likely to take a greater part in the arbitration process than they do now. Seemingly, the time has certainly not come for the AIs to take over yet. The current AI arbitrators’ ability to render a just and sufficient arbitral award all by themselves seems to be highly controversial and most domestic arbitration rules do not expressly permit non-human arbitrators to render awards.[10] Therefore, it seems appropriate to put aside the first option as a future goal and go with the second and third options to utilize AI arbitrators alongside human arbitrators and evaluate the reaction of different laws on the matter.
An Analysis from a Turkish Law Perspective
Although AIs are not acting as judges or arbitrators for now, in the near future, Turkish courts may be faced with recognition and enforcement requests for arbitral decisions rendered by AI. In that case, what will happen from a Turkish law perspective if an AI is chosen to act as an arbitrator?
According to the principle of freedom of contract, it can be claimed that the parties can insert a clause in their arbitration agreement stating that an AI arbitrator shall rule the case and render an award. However, the issue is the question of legal enforceability of that award, which brings along the two following points of significance that must be considered: (1) the validity of arbitration clauses regarding the appointment of an AI arbitrator between the parties and (2) the validity and enforceability of arbitral awards rendered by AI arbitrators.
The New York Convention dated 10 June 1958 (“NY Convention”), which the Republic of Turkey is a party to, does not specifically define an arbitrator as a human being.[11] The NY Convention does not hinder the will of the parties of an arbitration from choosing an AI arbitrator, nor does it regulate any such rules. Article V/2/b of the NY Convention emphasizes the significance of public policy while rendering such awards. Yet, “most domestic arbitration laws are silent on the use of AI arbitrators,”[12] including Turkish legislation. In line with the foregoing, the Turkish Code on International Private and Civil Procedure Law no. 5718[13], Article 62.1.b solidly establishes that an arbitral award is not enforceable in Turkish courts if it is disorderly to public morality or contradictory to public policy.[14] Likewise, Article 15 of the International Arbitration Code no. 4686[15] prohibits the enforceability of an award by two of its provisions: (1) 15.1.a states that any agreement that is null and void in the chosen law of application is subject to annulment and (2) 15.2.b states that any award is subject to annulment if it contradicts public policy. Article 7/B/1 of the same Code states that “Unless otherwise agreed to by the parties, the following rules apply in the selection of arbitrators:1. Only real persons can be chosen as an arbitrator.” Since an AI is not legally permitted to have any status of personality according to Turkish law, it is clear from the wording of the law that AIs are prohibited from acting in the capacity of an arbitrator unless otherwise agreed to by the parties.
Although the above provision seems to be illuminating with regard to the freedom of contract, the validity of such agreements is delimitated to certain rules as well. According to Article 27/1 of the Turkish Code of Obligations, agreements that are incongruent to public policy are null and void. Based on this provision, the answer to the question of whether the appointment of an AI to act as an arbitrator is congruent to public policy will determine the validity of such agreement. Therefore, public policy stands as a strong measure in Turkish law as it is considered by the courts to ex officio overrule the freedom of contract principle in determining the enforceability of an arbitral award.
As highly respected as it is, woefully, the scope of public policy remains to be an unregulated area and what constitutes a breach of public policy in Turkey is obscure. In consideration of the foregoing provisions of law and since a non-human making rulings regarding justice for humans would raise the bar too high in means of social order, a legitimate use of an AI arbitrator seems to remain a ways away. Hence, the rules regulated by the law display the will of the lawmaker in prohibiting a nonhuman subject interfering with the legal system. At this point, although AI tools may be used, public policy remains an obstacle for AI-regulated arbitrations and Turkey is not likely to have AIs replacing lawyers or arbitrators in the near future without explicitly overcoming such legislative obstacles.
The Future of AI in Legal Practice
In conclusion, while the utilization of AI in arbitration comes thrilling to many, no one has yet been able to reach a peak point of its use. In the present state, AI becomes a widespread tool assisting arbitrators and lawyers, as their algorithms are a more flawless and trustworthy tool used to eliminate the possible breaches and leaks of substantial information put into the system. While the time and cost effectiveness of AI is attracting more and more users around the world, using AI as a sole arbitrator still remains a vision many look up to, though it would come as a facilitation to many complicated procedures of arbitration. Although the current legal system and the rules in Turkey AI seem to be nothing more than a futuristic dream, there are some promising developments for the use of AI in arbitration. The Istanbul Arbitration Centre (ISTAC) is working on the use of AI in the preparation of arbitration clauses as well as the follow up of arbitration case files.[16] The exclusive use of AI in rendering arbitral decisions is not a goal for now, as countries value human touch at least for the predictable future. Turkish law may need some modifications that facilitate the enforceability and validity of AI-related awards in order to position itself as a proactive legislation in the arena of fast-developing technology of near future.
Lex Machina is a massive statistical tool that derives data from a catalogue of court cases and uses them to create statistics regarding the case at hand. If the tool has a similar case in its catalogue, it can estimate the given case’s likelihood of success and/or failure. See https://lexmachina.com/what-we-do/ ; Arbilex is a statistics-based prediction tool created specifically for arbitration procedures. It determines which of its case prediction models would be the most adequate one to predict the given case based on past data. See https://www.arbilex.co/welcome ; Arbitrator Research Tool (ART) is basically an international database of arbitrator information that contains and provides information on gender, nationality, experience, applicable laws, etc. and a list of arbitrators who competed against the arbitrator under research. See https://globalarbitrationreview.com/arbitrator-research-tool ; Liti-gate is a platform not dedicated solely to arbitration, but it manages to help in every area of litigation. It makes an analysis, crosschecks information, prepares timelines, provides responses and draft arguments for the case, creates bundles from a selected range of documents, and also deals with legal research. See https://www.litigate.ai/features/ ; Arbitrator Intelligence is another AI tool that is used specifically in arbitration, but unlike the other AI tools, it makes its analysis based on the questionnaire it provides for its users to fill out. Based on past data and according to the answers in the questionnaire, it makes predictions about the timing, duration, case management, document disclosure, rulings on jurisdiction and interim relief, damages analysis, legal methodologies, and also information on the arbitrator in order to plot a route for the case and provide a better evaluation on choosing the arbitrator(s). See https://arbitratorintelligence.com/about-1
