Introduction:
There are umpteen reasons why the arrival of an Artificially Intelligent Judge seems inevitable. Humans generate billions of disputes every year, soon to be tens of billions as the growth shows no signs of slowing. We cannot help ourselves; we love to fight with each other. Conflicts are universal and thus, there is an immense need to develop processes to minimise, manage and decide upon these issues in a manner more efficient than current human-based resolution. This is where technology comes into play. Technology has already enhanced dispute settlement through video conference hearing, electronic briefs and automated online processes, popularly known as ODR (Online Dispute Resolution). But can Artificial Intelligence be used to make the process even smoother and faster, and perhaps, can the machine become the arbitrator itself?
Many experts are of the opinion that the purpose of techniques of ADR such as Mediation, Negotiation, Arbitration etc, is to bring back the human element and to resolve the dispute in an informal, flexible way as opposed to the strict procedure of courtrooms. Bringing a machine into the picture to settle the dispute through a formal, emotionless and standardized procedure may seem counter-intuitive to its very purpose. However, recent innovations such as the ‘birth’ of Sophia are a harbinger of things to come. Sophia[1] is a robot with a humanlike form who is capable of socializing with people and building rapport with them! Given the rapid speed of developments, the day is not far where we would witness an AI lawyer capable of understanding human emotions, arguments, detecting lies and rendering decisions on the basis of data stored in it. This article seeks to analyse how such a technology would function in ADR, its benefits and limitations as well as potential to eventually replace lawyers!
How Would an AI System Function in the Field of Dispute Resolution?
The AI can work in 2 ways:
- AI can be a tool for the adjudicator
- AI can be the arbitrator or adjudicator itself
In the first approach – The AI can help in reviewing all the documents, researching similar cases and drafting standard clauses. It can also make predictions of the outcomes, calculate the amount of damages, detect lies and suggest probable solutions. Thus, human decision-makers would be consulting the AI on an advisory basis to greatly speed up and simplify the ADR proceedings.
In the second approach – Both parties can be asked to put forward their last, best offer and the algorithm would research in its database and see which proposal is closest to its model solution. This would also encourage the parties to put reasonable offers so that the AI picks their offer over that of the other party’s. This design plays to algorithmic strengths and avoids subjective questions that might trip it.
In a more advanced setting, a database of prior cases will be created containing more than ten thousand decisions made by adjudicators regarding a specific issue, ex- penalty payments or traffic court decisions. The details of each violation such as severity, scope, and type of information shared, facts would be stored in the database. The algorithm then crawls through all cases and creates a set of rules which correlate the decision rendered in each case to the details of each case. With this setup, whenever a new case is presented to the AI, it will consult the rules it created while learning from the raw data and will make a determination as to the appropriate payment amount or penalty to be imposed. Thus, when given accurate and reliable data, this principle can be extended to any number and category of cases.
Benefits of Using Artificial Intelligence in ADR
- Explaining the process – The simplest method of using an AI would be at the beginning of the session. The machine can answer questions and address doubts as to how the ADR process would unfold. A robotic machine can repeat the same information numerous times without growing impatient as a human arbitrator might.
- Time effective – One of the goals of ADR is to save time in comparison to litigation. Yet, lawyers are often burdened with endless documentation and research work, which is to be completed in a very short time frame. An AI can automate data analysis and research and hence, reduce the burden on the individual.
- Cost-effective – Although the initial set up would be huge, once the AI becomes operational it would be a cheaper alternative to traditional human-based resolution. Lawyers charge their fees per hour while the AI might require only operational and maintenance charges to be paid.
- Sharing sensitive details – Some people might be more comfortable revealing private details to a robot rather than a person. For example – In a mediation between two spouses who want a divorce. The parties might fear being criticized by another person or being blamed for the divorce. A robot cannot judge anyone which might make people open up to it.
- Documentation – ADR is a shorter process than litigation, which makes AI particularly helpful in reducing documentation workload. The document production process in litigation can go on for two years without anyone batting an eye and the same has to be completed within a few months in ADR. AI can help by quickly reviewing and selecting the material documents, or for preparing summaries of the documents.
- Drafting – Arbitrators spend a large portion of time drafting standard sections of arbitration awards, such as – the parties, procedural history, the arbitration clause, the governing law, the party’s positions and the arbitration costs. This work of drafting ‘boilerplate’ sections may be delegated to AI machines thus, freeing the Arbitrator for more creative tasks and saving the time and fees of the parties.
- Recording – AI may render the use of court reporters obsolete as it is capable of recording hearings, converting speech to text and providing a real-time transcript with speaker identification.
- No scope for bias or unconscious control – The decisions made by an AI system would not be tainted by human weaknesses such as prejudice, bias, illogicality or simply having a bad day or being tired. It also takes care of other human tendencies, such as relying on the first piece of information received, or being affected by the external environment or the other cases the arbitrator dealt with earlier the same day.
- No conflict of interest – When robots are appointed instead of human arbitrators, there would be no grounds for challenging the arbitrators on account of bias or a conflict of interest.
- No scope for errors – Human Arbitrators can make mistakes in interpretation, translation, documentation, authority selection, decision making, etc. Using AI at various stages or in different tasks can help to eliminate the inefficiencies in the arbitration process.
- Outcome forecast – AI can be used to select the appropriate arbitrators and predict the outcome based on the information given, the documents submitted, and the arbitrator’s reasoning. It can also be used to predict what would happen if the dispute goes for litigation, the range of damages and the most likely solution reached, thus encouraging the parties to settle their conflict through inexpensive ADR only rather than letting it reach the courts.
- Immediate Award Creation – Parties usually have to wait for compliance of the award after it is passed. AI will allow for an award to be implemented immediately. For example, in the event of a dispute over money, an award is passed that one party (A) must give a certain amount of money to another (B), then using AI, the sum is transferred immediately from A’s bank account to B’s bank account. AI can also give a recall to the concerned party and authorities at regular intervals for the compliance of the award.
- Detecting lies – While humans may not always be able to understand the body language of another person, an AI can be trained to monitor heart rate, blood pressure, eye movements etc. Thus, it could serve as a lie-detector during the proceedings and would help advance the ends of justice.
- Access to justice – The use of AI will lead to greater transparency in the laws, more reliable and quickly available legal information, time and cost-effectiveness. The technology will help in addressing the challenges surrounding justice and affordability and will thus, greatly improve the access to justice.
Challenges in the implementation of AI in ADR
- Enormous investment – While one of the main goals of an AI would be to reduce the cost of proceedings, a huge amount of initial capital is required to undertake research and testing of such AI. The more intelligent the program is, the higher would be the cost to develop it.
- Replacing lawyers – It is a frightening prospect for legal professionals that the technology which was initially developed to help them, may one day take away their jobs! There would only be a requirement for a few people to make the AI system function and the employment rate of mediators, arbitrators and the like would drastically decrease. In 2016, a report by Deloitte UK estimated that 39% per cent of jobs in the legal sector may be automated in the next ten years and that the industry will undergo profound reforms[2]
- Privacy concerns – Confidentiality of the information shared is a pivotal point in ADR proceedings. While a human can be trusted not to divulge the information, miscreants with the required skills can hack the AI software and access the confidential data. The persons involved in programming the software may also have full access to the same. Such a situation would warrant the functioning of the AI to be regulated by data protection and privacy laws.[3]
- System vulnerability – Like all technological inventions, the AI system would also be prone to hacking, spamming, viruses, software malfunction and other computer vulnerabilities and technical problems.
- Lack of flexibility – An AI would make use of data from prior proceedings and apply the general law of the state, without taking into consideration the subjective aspects, emotions, motives etc of the parties. Every case is not the same and occasionally, arbitrators may choose to divert from their previous decisions. This is not the case with an AI as it would have a standardized process for determining the type of case and would render the decision based on a fixed algorithm. The decisions would be conservative and there would be no scope for development. Such a rigid structure would lead to only a few combinations of judgments existing and might even bring the gradual evolution of Arbitration and Mediation laws to a standstill.
- Legislation required – The technology would be a revolutionary change in the world of ADR and would require to be accompanied by a legal framework. The Arbitration & Conciliation Act, UNCITRAL rules, procedural and institutional laws will have to be amended. Several questions would have to be decided such as – Who would have access to the system? Would it function independently or be monitored by a human? What would be the cost? Would there be a system of appeals to a more intelligent AI?!
- Ethical considerations – All jurisdictions across the world have provisions for a fair trial. This implies the presence of a human as they combine the strict rules of law with the principles of equity and good conscience. A robot would not have any awareness of these subtle concepts.
- Lack of explainability – Developers may refuse to explain how their program works or might themselves not be able to understand how the computer reached a particular prediction or decision. It is a principle of natural justice that legal decisions must compulsorily be supported by a rationale or reasoning. Some regulations such as the GDPR of the European Union have even established a “right to explanation.”[4]
- Failure to consider the option – It is possible that despite the heavy amounts spent developing the software, people would be wary of using it. Many would not be willing to let a machine dictate their life decisions nor would they have the same respect for it, as they would for a decision taken by an actual Judge. It is also possible that parties who tried the system earlier lose faith in the concept of a Digital Judge due to the lack of flexibility, accountability, threat of confidentiality and other technical complications which they may have experienced.
- Lack of human touch – People might prefer to have a human Judge who can be convinced by using emotional appeal or narrating their case in an effective manner. Is there any way to convince a machine once it has made up its ‘mind’? Further, would a computer know the right questions to ask? There can be hundreds of varying circumstances in ADR. Humans are skilled negotiators and experts in knowing which questions to ask, how to probe into an issue etc. While an AI may have certain uses, there are certain tasks in ADR which primarily revolve around human understanding and communication and cannot be completely replaced.
Present Scenario and Recent Developments
In February 2019, a robot mediator settled a dispute for the first time. ICan systems famously became the first company to resolve a dispute in a public Court in England and Wales by making use of this AI.[5] The AI tool is known as Smartsettle ONE. It replaced a human mediator and in less than an hour, settled a three-month-long dispute over a £2,000 unpaid bill by using a blind-bid kind of mechanism!
How it works – The system allows parties to make offers and counter-offers by moving flags along sliders. The algorithm learns the bidding tactics and priorities of the parties and ‘nudges’ them into a settlement, without revealing their secret bids. This mechanism helps parties avoid the negotiation dance where they put forward offers which are not their best. Further, it rewards the party which makes the greatest effort to settle.
Graham Ross, a lawyer and mediator who has used this system has argued that “Mediators may complain that this will put them out of business. But the truth is these systems empower the mediator to be more effective by improving the chances of the parties reaching an agreement. There’s nothing to be lost by using it — it will either settle or it won’t.”[6]
Currently, a newer version called Smartsettle Infinity is being designed, which allows parties to build conditions into the process, and is particularly suited to business disputes. Conditions could include an agreement to withdraw a bad consumer review, continue buying products from a company or pay by instalments.
Post COVID Impact
The pandemic has accelerated the trend towards using smart technologies in the legal landscape. E-courts are conducting virtual hearings, online dispute resolution is on the rise and real-time screen sharing, evidence presenting programs are being used. With this increased focus on technological adaptations of traditional methods, the day is not far when we will witness completely automated procedures with the use of AI systems.
Conclusion
Would claimants tolerate their cases being resolved by computers? Many consider justice to be a uniquely human ideal that may be extremely difficult to execute in Artificial Intelligence regardless of its level of sophistication in programming. There is a school of thought that people are less likely to trust a decision made by a computer (even if it is based on clear logic) and may easily insist that the computer must be malfunctioning if they do not get their preferred decision. However, the countervailing argument is that AI is increasingly becoming a part of our day-to-day lives – to the point that we allow it to drive us and our families around in self-driving cars. There will come a time where we are completely comfortable in letting the algorithm adjudicate our case for us.
Technological advancements are being made rapidly and we cannot avoid its integration in legal services. Thus, the best policy would be to embrace this change and tactically utilize it for the advancement of the legal profession. At its current stage of development, the efficacy of AI is highly dependent on the quality of data processed and the algorithm applied.[7] The focus must be on creating the right mix of human-machine capabilities to re-cast future relationships that benefit both provider and clients. It is important to remember that Artificial intelligence is not aimed at taking over the legal profession, neither is it a threat to retrench lawyers. Rather, AI is focused on making the job of lawyers easier, faster, more productive and to provide efficient legal service to all.[8] Janet Fuhrer[9] has remarked – “Even with technology, certainly there are still lots of opportunities for lawyers to provide services.” While this leap from human-powered justice to electronic justice is a big one, much like the concept of self-driving cars or phones we can talk to, many people seem to have concluded that this future is inevitable. Right now, we’re just biding time, waiting for the future to arrive.
References:
[1] Zara Stone, Everything You Need To Know About Sophia, The World’s First Robot Citizen Forbes (2017), https://www.forbes.com/sites/zarastone/2017/11/07/everything-you-need-to-know-about-sophia-the-worlds-first-robot-citizen/#404eb33446fa (last visited Sep 3, 2020).
[2] Andrew Yang, THE WAR ON NORMAL PEOPLE: THE TRUTH ABOUT AMERICA’S DISAPPEARING JOBS AND WHY UNIVERSAL BASIC INCOME IS OUR FUTURE, 38 (2018).
[3] Mayank Samuel, Confidentiality in International Commercial Arbitration: Bedrock or Window-Dressing? Kluwer Arbitration Blog (2017), http://arbitrationblog.kluwerarbitration.com/2017/02/21/confidentiality-international-commercial-arbitration-be. (last visited Sep 3, 2020).
[4] Sandra Wachter, Brent Mittelstadt & Luciano Floridi, Transparent, explainable, and accountable AI for robotics, 2 Science Robotics (2017).
[5] Ian Burns, Settlement of legal dispute using B.C.-produced robot mediator called first ever – The Lawyer’s Daily www.thelawyersdaily.ca (2019), https://www.thelawyersdaily.ca/articles/11621/settlement-of-legal-dispute-using-b-c-produced-robot-mediator-called-first-ever (last visited Sep 3, 2020).
[6] Ibid.
[7] Edna Sussman & Kathleen Paisley, What Do We Mean by Artificial Intelligence? 6 (2018), https://sussmanadr.com/wp-content/uploads/2018/12/artificial-intelligence-in-arbitration-NYSBA-spring-2018-Sussman.pdf. (last visited Sep 3, 2020).
[8] Carl Wiens, Law and Technology: How will artificial intelligence affect the legal profession in the next decade?, http://law.queensu.ca/how-will-artificial-intelligence-affeect-legal-profession-next-decade.
[9] partner with Ridout & Maybee LLP, Ottawa, and President of the Canadian Bar Association.
0 Comments