NEWSLETTER

Arbitration with AI

Introduction-
The three general pillars of arbitration can be classified as namely, (i) speedy, inexpensive
and fair trial by an impartial tribunal; (ii) party autonomy; and (iii) minimum court
intervention. Holding these principles at the core, the exploration of generative AI models for
arbitral and judicial purposes is a welcome change and could have highly positive yet
unfathomable results. With law being open to interpretation and having to have wide scopes,
the prospects are immense- the currents of technology and law are seemingly colliding and
could work parallelly in the future however, there are concerns pertaining to their ambit and
furtherance. One of the prevailing concepts being that of Online Dispute Resolution
surprisingly dates back to 2001, wherein Profs. Ethan Katsh and Janet Rifkin authored ‘Law
in a Digital World’, actively introducing the possibilities and probabilities of the said field.
Change is slowly yet surely happening and must be accepted.

Machine Learning and Artificial Intelligence (ML/AI)-

The inception of ML dates back to the 1950s with Alan Turing considered as the Father of the
field. The core difference between ML and AI is the independent ability to think and reason.
ML has limitations due to its shortcomings; the reliance on human supervision, training and
input. The main purpose served is saving time. ML essentially functions on blockchains
whereby computers/ programs are trained to perform specific tasks using algorithms and
data- the process, outcome and code are then saved in the form of an extensive network of
easily accessible databases for human participants as well as machines. The information
stored is available to the participants of that specific blockchain which is an online cloud
however, they cannot make changes to the functioning and code of the blockchain.
With arbitration, ML has introduced supervised learning wherein input data is appropriately
paired with output data to identify patterns and streamline the relevant data with respect to
the task i.e. the legal dispute inputted. This has mainly led to 5 developments which can be
classified as secure case management systems, speedier research and its compilation,
summarization and identification of problems, strategy development and predictions of
outcomes. Singapore and China have made noteworthy developments with the Singapore
International Arbitration Centre (SIAC) having introduced a draft of its seventh edition of the
SIAC Rules, which includes provisions for the SIAC Gateway; an online case management
system. Drawing inspiration in 2022, the ICC introduced "Case Connect," in line which is a
secure online case management system that enables arbitrators, parties, and the Secretariat to
communicate and share documents efficiently. This strongly expedites and streamlines the
arbitration process by solely leveraging technology.
Additionally, China is exploring AI with law by testing programmed AI assistants to draft
decisions, with institutions like the Shenzhen Court of International Arbitration exploring
AI’s potential to help arbitrators analyze cases and review documents. Tools like ChatGPT,
Jus Mundi’s AI-powered research assistant, and Thomson Reuters’ co-counsel are already
assisting arbitrators and lawyers in tasks ranging from legal research to document
summarization. Firms are automating the comparatively more mundane tasks which require
no creativity or generation of code instead of the mere running of code- such as compilation
of data and research as well as the identification of likely loopholes and counters. Frivolous
claims and probabilities of the issue are also being analysed and identified by these
organisations’ AI models. Parties are therefore, allocating funding towards levelling,
democratizing and stabilizing the market, resulting in smaller firms competing with bigger
firms due to their focus on substantive issues. Evidently, this aligns with arbitrations’
principle of expedited solutions; parties are increasingly favouring this collaborative system
of ML/AI for their arbitration disputes due to its’ almost instantaneous and automated/online
procedure. With respect to India, the Ministry of Law and Justice has also promised
exploratory research on the implementation of ML/AI in the judicial and arbitration system.
This presents a lot of hope due to the sheer number of pending cases in the country.
While Online Dispute Resolution (ODR) arose from the completion of the entirety of the
proceedings online, subject to exceptions, it has now increased its’ ambit and capability. The
medium of conduction used to be a specific online platform however, the tasks were not
automated with the process and proceedings being identical yet different in the mere sense of
mode of submission of documents and the hearings being online. E-evidence had also been
getting preference with the arbitration panels accepting it. Arbitration Courts, organizations
and firms have started expanding their exploratory radar; WIPO, SIAC, UNICTRAL etc.
have promised developments with respect to ML’s incorporation in e-commerce arbitration
and online consumer mediation. Up until this point, ML is being used for blockchain
verification, inputted data’s research, compilation, strategy development and probability
check. However, how far can analysis take us? Advancements in AI are now leading to their
usage in judicial and arbitral proceedings which could be a cause for concern; the legality of
arbitration being open-ended and positively disposed with arbitration’s purpose and
proceedings may result in problems.

Noteworthy Issues-
1. Minimum court intervention-
In case of the wide-spread application of Generative AI, judges may not have to be trained in
law due to the judicial assistance provided by AI. Small claim tribunals in Singapore have
started experimenting with AI driven justice wherein the bot analyses the facts, refers to
relevant precedents and delivers a verdict in accordance. However, the parties are asked of
their satisfaction, with the case being transferred to a human judge presiding over the AI bots
courtroom, in case of party grievances. This could extend into the proceedings of arbitration
with high cost-effectiveness and expedited hearings yet there is a lack of reliability and
transparency that seeps in; this could postulate problems, especially if applied to Arbitral
Awards. It is bound to eventually reach Arbitral proceedings with time yet the question on
humans trusting AI to administer justice, even if the process is efficient is a looming shadow.
 2. Common concerns with AI-
The advent of AI created numerous concerns in the industry- namely surrounding reliability,
accountability, confidentiality and the lack of transparency. The same would be prevalent in
the field of Arbitration, if inculcated. The solutions and answers provided by GPT are still
unverifiable and raise a question of reliability with a possibility of increased error with
respect to Arbitral awards- in way meeting time constraints and concise implementable
verdicts. There is a lack of transparency in the process as other than the creators and coders of
the system; the operators and users do not have access to the code which results in them not
knowing the workings and procedure of the system. Arbitration proceedings are often
preferred due to the transparency they offer however, parties may not opt for ODR due to
hesitancy and ambiguity on how the Arbitral Award is reached, formulated and passed.
Accountability and confidentiality of the systems will naturally fall upon the operators i.e. the
Arbitration firms or the international/national Arbitration bodies. However, who should be
held answerable in essence? If an automated system errors, do the creators incur liability or
the owners?

3. E-evidence-
In the possibility of a fully functional online arbitral proceeding i.e. ODR- the evidence
submitted would devolve into electronic means naturally. Humans/ client-parties will have a
lack of confidence in the same with human arbitrators facing difficulties in verification of the
proof submitted; new and uniform guidelines will have to be issued. The development of
faith in the procedure requires a passage of time, with natural errors along the way.
Additionally, a rising issue is that of “deep-fake” testimonies with people creating e-evidence
on their devices and pre-recorded testimonies with the help of generative AI and technology
which have close to no means of verification. The ease of Court and evidence
manipulation/tampering will heighten with AI’s advent and incorporation into the arbitration
system- the fundamentals of the process will be put at risk. Relevant guidelines and rules can
be put in place to tackle the same.

Probable future-
Contrary to rising fears, the possibility of a complete take-over of AI on the judicial system
seems highly improbable owing to the slow developments of ML/AI, while also having
severe limitations on scope. The ability of AI developing rationality and independent thought
is a far-off and long drawn idea if possible, at all and would therefore, be unlikely to be
entrusted with the task of passing an Arbitral Award and/or judicial judgement that has real
consequences. While the responsibilities could extend to an assistor’s capacity with AI
revolutionizing cost-effectiveness and expedited arbitral proceedings along with heightening
case management and strategy development for arbitration firms; humans would still be
needed to supervise but more importantly, to make creative decisions, come up with new
landmark judgements and lines of thinking as well as building human trust and interpersonal
relations with clients and parties. Essentially, generative AI will not develop soon, if at all- to
have a mind of its own. However, the incorporation of ML in Arbitration in the near future
will lead to positive outcomes. Legal jobs will be retained, if not created with an addition of
roles. The playing field for arbitration firms will be levelled with smaller firms competing
with the bigger firms due to stabilizing and equalizing tools i.e. analytic ML/AI. Arbitral
proceedings would be expedited and deliver to the client/parties’ satisfaction with frivolous
claims cleared up on the basis of prediction by the Arbitration-counsel firm. The tedious and
bothersome paperwork and research will be taken off the hands of young interns and
paralegals and instead be merely inputted to the ML program, subject to verification.
Blockchain based proceedings will be efficient, expedited and cost effective if explored and
consequently implemented. The future of arbitration with respect to revolutionized Online
Dispute Resolution and ML’s assistance seems assured and positive- onto a new era.

Best,
Ruhika Kohli
III- [Link].B.
Arbitration Cell, ILSCA