The term Artificial Intelligence itself explains that technology having human capabilities of performing tasks that require human intelligence.
This topic is intuitively interesting that one day our disorderly, frenzied human disputes will be determined by the calm, all-knowing logic of a fair and neutral electronic decision maker. While the concept may be appealing, this jump from human-powered justice to electronic justice is huge.
Much like the concept of cleaning robots or self-driving cars, many people seem to have deduced that this future is inevitable, even when we don’t yet have a full developed technology that could take over. Currently, we are just anticipating the changes that will unveil with time. There are several reasons that the arrival of the digital judge is unavoidable. We humans create billions of disputes each year, soon to be tens of billions. This escalation of disputes shows no signs of rest. Humans want of fixing problems instantly is creating burden on the system. The idea that present load, human-based resolution processes could resolve all these disputes poses a serious question.
Faith in our very capability to be fair and unbiased arbitrators dwindles under this strain. Alongside these developments, computers carry on to become more dominant and more deeply integrated into our daily lives. It stands to reason, then, that if current trends continue, computers will one day be better at justly resolving our disputes than we are.
In order to understand the use of AI in Arbitration, one has to understand how AI is used in different areas. AI, in the form of robotic technology, is used in the field of psychology for achieving therapeutic results. Robots are used for dealing with the child’s psychological condition by programming it to elicit positive and desirable responses.
Psychology, despite being a complex and sensitive field, has such a wide acceptance of AI, therefore, incorporating AI in the field of Arbitration is possible and can elicit productive results. The AI-enabled systems can be used in the dispute resolution process in various forms.
The whole Arbitral Tribunal can be replaced by an AI-enabled system. This will eliminate the process of selecting the Arbitrator by the parties or appointment of a sole arbitrator by the Court. Although the result in this process will be solely based on the algorithms used in the program of the system.
Hence, the power of deciding the case will be in the hands of the programmer. This would be the most controversial issue and there are chances of non-acceptance of the award as the right to choose the arbitrator will be removed.
Further, if the decision is solely based on the algorithms, then the purpose of providing distributive and equitable justice will be defeated. As the human arbitrator considers various factor, such as the income and condition of the parties, before passing an award which might not be considered while using the AI-enabled system in the decision-making process.
Hence, there might be more practical approach by human arbitrators than the AI-enabled systems while passing an award.
Appointing Human Arbitrator along with an AI-enabled system can help in reaching to a conclusion much faster. AI can manage the analysed documents before the award is passed and answer the queries asked by the arbitrator while passing the final award.
It can act as a tribunal secretary. But there can be some obstacles or disagreements, for eg. the human arbitrator ignoring the logical answers provided by the AI-enabled system after analysing the evidences, which may result in a delay in the process discouraging the purpose of Arbitration.
AI to provide an expert opinion to the human arbitrators The AI-enabled system can be programmed in such a manner that it provides an expert opinion based on the evidences furnished in the proceeding of the arbitration. There are more chances of acceptance by the parties as the final award will be passed by the human arbitrator.
> Researching, Summarising and Analysing the arguments and pieces of evidence
The AI-enabled systems can prove to be efficient in analyzing and summarizing large documents, databases, etc. in much less time than humans. By using these summarised information and analysed evidences, responses can be generated, through in-built algorithms, for solving the issues and reaching to a conclusion effectively and efficiently.
A similar kind of AI is used by IBM named ‘Ross’. Ross is efficient enough to read and understand a large database in less time. When questions are asked it can postulate hypotheses, compile the relevant research and generate responses accordingly along with citations. Ross is a kind of Strong AI that develops and learns from its experience. There can be wide acceptance of using this kind of AI in the field of Arbitration for processing, summarising huge cases, documents, commentaries and clarifications of the queries.
The appointment of a person who transcripts the hearing in front of the Arbitral Tribunal involves an additional cost, high chances of errors and it becomes a time-consuming process. AI can be helpful in such transcriptions as the machine might be able to record the hearing via microphones and provide an immediate transcript of the hearing.
> AI’s decision v. Tribunal’s decision
The decision provided by the AI-enabled system after analyzing the relevant piece of information can be compared to the decision arrived by the Arbitral Tribunal.
The decision of AI will either be complementary or contradictory to the decision of the Tribunal. This will enhance the reasoning of the Tribunal and the award will be passed after considering all the factors. Therefore, AI’s decision can be taken into consideration to remove or correct the unconscious biases of the arbitrators.
> Selecting an Appropriate Arbitrator
AI can aid in selecting an appropriate arbitrator for a particular kind of dispute after analyzing the record of the arbitrators and which arbitrator has experience in dealing with particular kinds of dispute.
This will reduce the burden of the parties (or the court in case of non-appointment of the arbitrator). The chances of equitable justice will be high when an experienced arbitrator will deal with the dispute.
> Drafting suggestions
AI technology can be inculcated for getting suggestions on the drafting of an Award passed by the arbitrator in an appropriate manner. The award passed has to be in a meticulous manner and using AI for this purpose will reduce the time taken for drafting the award, hence serving the primary objective of arbitration.
> Natural Language Processing
Natural language processing (hereinafter NLP) refers to automatic computation and processing of human languages and gestures including their speech, texts, documents, commentaries, digests, made by them using their creative thinking abilities.
The use of an AI-enabled system as NLP can be done to analyze and extract meaning from huge and complicated documents that may be relevant. For Eg, DiaLaw- a two-player dialogue game.
The statements and arguments of the parties are recorded in the system and there is a bifurcation of the relevant information required for the dispute to be resolved. This is a rigid tool as the language used is formal. It helps in getting a clear picture of relevant arguments. The game ends when either party accepts the statements of the opposite party or rejects it. This is a tool for supporting NLP.
Further, the AI-enabled technology as NLP can also be used to translate the documents which are in a different language than the one used in proceedings, including scanned, handwritten and printed documents in less time with a high level of accuracy.
There are 2 kinds of AI- Strong and Weak. Weak AI mimics the human abilities to think and reason. It doesn’t learn by itself and only uses the pre-programmed human abilities for functioning. On the other hand, Strong AI can think and reason like humans independently.
They do use pre-programmed human abilities but for creating their reasoning. Weak AI cannot be used in the legal field it is ultimately in the hand of the programmer, what the AI will think and give reason accordingly.
Strong AI can be advantageous to the legal field as it will reduce the load off the people working to resolve the case.
Following are the benefits of using AI in the field of law, especially Arbitration:
Humans have 3 parts of the mind: conscious, subconscious and unconscious. The conscious part of the mind is the one that we use for making a decision but the subconscious and unconscious minds have also a role in our decision-making process. It is expected that an Arbitrator have a rational approach while delivering the decision and award but there are chances of having an influence of the external environment over human thinking and reasoning.
All the 3 parts of the mind have an influential effect while making decisions. For e.g., while passing an award if party tries to approach the arbitrator emotionally, it impacts directly to the subconscious mind and there are chances that a lenient award is decided. This hampers the rational approach of humans.
AI programs are not influenced by any such unfounded factors. They are programmed to act independently using algorithms. Hence, the decision made using AI is much rational than humans.
Humans are affected by cognitive biases. There can be instances that the arbitrator might be influenced by the cases he dealt with earlier on the same day or the external environment while deciding the case in the evening.
Their approach in decision making can be affected. An example of cognitive biases is the anchor effect, where there is a common human tendency to rely on the first piece of information received to make subsequent decisions. Humans have a tendency to such cognitive biases while approaching a decision.
AI programs do not have an effect of the external environment and they do not suffer from cognitive biases.
The primary purpose of arbitration is to achieve a fair resolution of disputes by an impartial third party without unnecessary expenses or delay. Saving time and cost are the core aims of Arbitration process.
The use of AI in the arbitration proceedings will be a supplement to the objective of arbitration. Since AI is developed to think and reason independently, it can be used to reduce the burden of the person involved in the proceeding of arbitration. The use of AI will automate the time-consuming labour of legal research and data analytics.
Human Arbitrators can make errors in interpretation, translation, documentation, the appointment of authority, decision-making, etc. The use of AI at different stages or in different tasks can help in eliminating the inefficiencies in the arbitration proceedings.
It can identify blind spots and recommend suggestions to alleviate such blind spots in order to make the process efficient. AI-infused products and services could assist in better management of cases by detecting inefficiencies and automating management tasks.
AI can be used to choose the appropriate arbitrators and predict the result based on the information given, documents submitted and the reasoning provided by the arbitrator.
In the present scenario, after passing of award, the parties have to wait for the enforcement of the award. There is a gap between the pronouncement and enforcement of an appeal or application is not made for setting aside the award or for a stay on enforcement.
AI will enable the immediate enforcement of an award. For eg., In case of dispute in respect of money, award is passed that one party (A) will give a certain amount of money to another (B), then using AI, the amount is immediately transferred from A’s bank account to B’s bank account. AI can also send reminder at regular intervals to the concerned party and authorities for the enforcement of the award.
AI systems are trained and operated with data sets. AI has the potential to affect the purpose of Arbitration negatively.
Initial Incorporation of AI in Arbitration will require a huge investment and time because adjusting with such dynamic technology requires training. Development of AI systems is done to reduce the cost of the proceedings but for developing such AI programs and intelligent algorithms huge amount of capital is required which will automatically increase the price of such a system. If incorporated, it will prove to be cost-efficient in the near future for the persons involved in arbitration.
The purpose of the development and use of AI is to reduce the human burden. But this has a direct impact on the employment rate as for making the AI system function, there will be a requirement of a few persons only. Moreover, AI will be competent to do the work done by humans resulting in a reduction of the workforce.
Confidentiality is one of the fundamental necessities in the Arbitration Proceedings. AI is completely based on the intellectual algorithms and software programming developed by a programmer and selected few have full access to those algorithms which might provide the final decision in the particular case. Software programming is prone to hacking.
There is a threat that the confidential information of the parties can be compromised through hacking. Any updating of the system also poses threat of virus and other complex technical issues.
Due to lack of reasoning, lack of transparency, a threat to confidentiality, and other such complications, the parties may not rely on the conclusion arrived at by the systems using the algorithms they will lose faith in the system.
There was a case in the US, commonly known as the ‘Loomis case’, where the conviction of Eric Loomis was based on the outcomes of the closed -source risk assessment software called COMPAS (Correctional Offender Management Profiling of Alternative Sanctions).
The software uses proprietary algorithms that had 137- item Questionnaires which was developed by a Private Company called Equivalent. An appeal was made challenging the conviction based on secretive algorithms that cannot be examined. This shows the non-acceptance of AI in the legal field.
There may be a threat that the programmers may not disclose the algorithms and how the system has reached at a particular conclusion. To completely rely on a non-human for adjudication is a tricky situation.
Every case in the arbitration is not the same, so if the decisions are made by system functioning of ratio and an organized method for deciding every case is same then it depicts the lack of flexibility.
Every award is passed stating the reasons that helped in reaching the conclusion. If, there are just few fixed algorithms then there will be just few combinations of judgments, thereby a rigid structure will be devised.
AI is created by programmers who are humans. While developing such intelligent algorithms there is a possibility of human errors. If such errors by programmers enter into AI application’s code, there are chances of erroneous conclusion and other inaccuracies.
Due to such human errors, the AI-enable system which was made for detecting and removing errors would result in additional errors, which would create an additional burden and cost.
The growing trend of people moving to outside court settlement for their dispute resolution itself encourages inculcating AI in Arbitration. The primary purpose of arbitration is to arrive at a solution in less time and to avoid the complexity of the court. Arbitration is ready to lead the way in adopting new technological and procedural innovations. Arbitral institutions, tribunals and practitioners have a responsibility to do so, particularly where such innovations can reduce cost and improve efficiency.
Presently, there is no legislation which talks about having an AI-enabled system at different stages of Arbitration. If AI-enabled systems are incorporated in the arbitration proceedings, there will be a requirement of some regulations to avoid any threats to the parties. This is a delicate area where the strengths of AI is to be used without jeopardizing core values of arbitration.
Irrespective we like it or not, artificial intelligence is going to play a chief role in international arbitration in the near future. The amounts at stake are massive and the advantages from artificial intelligence are too enormous to avoid it. Al has considerable potential benefits for international arbitration, but as members of the arbitration community we also need to understand the perils of AI, at what cost, and how this might influence international arbitration in general ways that may not be obvious.