Introduction
Artificial Intelligence (AI) has several definitions
due to its developing nature. AI, as defined by Professor John McCarthy, the
computer scientist who defined AI tools in 1955, is “the science and
engineering of making intelligent machines”. Also, AI is defined by the EU
Artificial Intelligence Act, proposed on 21 April 2021 and adopted in 2024 (the
“AI Act”), refers to systems that, for a given set of objectives, generate
outputs such as content, predictions, recommendations, or decisions influencing
the environments they interact with. Over the last few decades, there have been
phenomenal advancements in the use of AI in the legal field, especially in
International Arbitration. The inclusion of AI tools in Arbitration is
increasing and taking several forms and tasks, especially in today’s modern
society, which has been accelerated by the COVID-19 pandemic. There are AI
systems, for example, used to analyze arbitration clauses, along with systems
used to draft an arbitration clause. The online "Clause-Builder" tool
from the American Arbitration Association (AAA), a known AI tool, was created
to help draft arbitration provisions based on the facts of disputes as well as
the user preferences for the arbitration agreement's components. Moreover,
there are AI tools used for choosing suitable experts, arbitrators, and
counsel, a best-known example is Arbitrator Intelligence, which provides AI
reports on arbitrators' track records enabling parties and counsels to make
more predictable arbitrator selections.
Time has shown,
especially after the COVID-19 pandemic, that no profession is immune to AI
taking control, not even an arbitrator role. By enabling computer programs to
process material similarly to arbitrators, AI poses a challenge to the
international arbitration field. The development of technology and the rapid
use of new technological tools have led to several questions in International
Arbitration. One of the most challenging questions is whether Artificial
Intelligence Arbitrators (defined as AI Arbitrators) can take the role of human
arbitrators or are they just used to support arbitrators and increase
efficiency? The subject of whether AI arbitrators can take the place of human
arbitrators is complex. In reality, it pertains to a larger debate over whether
AI can serve as a decision-making authority.
Scholars such as Marrow, Karol, and Kuyan framed the
essential debate early by asking whether computers could truly serve as
arbitrators, a question they summarized as 'Are we there yet?' Their work
illustrates the long-standing uncertainty about when, or if, AI can function as
a decision-maker in arbitration.[2]
Similarly, the scholar Colorado emphasizes that AI’s influence is not limited
to efficiency gains; it could reshape the very structure of international
arbitration in the long run, raising fundamental questions about legitimacy and
trust.[3]
The usage of AI technological tools in International
Arbitration and its impact on the future of international arbitration make this
question crucial, even though there isn't yet a system created for an AI
arbitrator to replace human arbitrators and serve as the adjudicatory
authority. As a result of this ongoing debate, it is necessary to answer this
question by analyzing the increased use of AI in arbitration, along with the
legal documents and instruments, to determine what criteria arbitrators must
meet. In other words, to analyze and assess the applicable instruments,
conventions, regulations, and public policies on the criteria needed to be met
by an arbitrator and the recognition and validity of the arbitral awards issued
by the arbitrator from a legal-normative perspective.
This will include the analysis and assessment of
several legislations, conventions, and instruments used by arbitration
organizations, including the UNCITRAL Model Law on International Commercial
Arbitration and the New York Convention on the Recognition and Enforcement of
Foreign Arbitral Awards, along with the EU Artificial Intelligence Act, enacted
in 2024.
Moreover, this study will also follow a
social-empirical perspective, by analyzing the available statistical studies
along with analyzing a survey conducted by the author to conclude whether AI
tools are used in a way that will allow them to take the role of human
arbitrators or are just used to support arbitrators and increase efficiency. In
other words, to conclude whether AI Arbitrators can replace human arbitrators
from a legal and social perspective.
Methodology
This
study adopts a mixed-methods approach, integrating normative legal analysis
with empirical data obtained from a targeted survey. The survey was designed
for arbitration practitioners from multiple jurisdictions, identified through
sampling based on professional networks, conference participation, and
membership in recognized arbitration bodies. A total of 29 practitioners
provided complete responses and took part in the study, representing diverse
legal systems and professional roles. The analysis focused on six main survey
questions, exploring the frequency of AI tool usage, the types of AI
applications employed, the perceived advantages of these tools, and opinions on
whether AI could replace human arbitrators. The survey questions analyzed in
this study are available in the Data and Information section of this article.
Quantitative data were analyzed using descriptive statistics, while qualitative
responses were thematically coded to reveal common patterns and points of
divergence. Ethical standards were strictly observed, with all participants
providing informed consent, responses anonymized for confidentiality, and no
personally identifying information collected or stored.
1.
The Framework of the
Study
1.1
The Legal Context
The significant benefit
of international arbitration is the party autonomy concept. Party autonomy in
international arbitration, across Interstate, International Investment, and
International Commercial Arbitration, enables the parties’ option to decide on
several parts of the arbitral process, one of which is the selection of
arbitrators. This concept is what makes arbitration unique and different than
national courts, since in arbitration, parties may choose the arbitrators to
hear their cases based on their qualifications, experience, and areas of
practice.
However, the main
question is whether this autonomy extends to allow parties to choose AI
arbitrators instead of human arbitrators. In other words, we should check if AI
arbitrators are permitted to be appointed as arbitrators. This ongoing question
requires the analysis of the relevant applicable regulations and laws at both
national and international levels.
At the national level[4],
there are several countries, for example, the Scottish Arbitration Rules (Rule
3)[5],
and the Swedish Arbitration Act (Section 7)[6],
that expressly provide that arbitrators must be natural persons. In addition to
that, the French Code of Civil Procedure states that arbitrators should be
natural persons capable of carrying out responsibilities.[7]
Similarly, some national
regulations refer to arbitrators using gender pronouns. A known example is the
United States Federal Act (Section 5).[8]
From such use of pronouns, we can deduce that the arbitrators they are
referring to are natural persons, although it is not stated clearly that AI
arbitrators can’t be appointed as arbitrators, giving the possibility for AI
arbitrators to take the role of human arbitrators unless new regulations are
issued in this regard.
In the Arab world,
arbitration laws also reflect a strong preference for human arbitrators. For
instance, in the United Arab Emirates (UAE), while the Federal Arbitration Law
No. 6 of 2018 concerning arbitration was enacted to align domestic arbitration
with international best practices, it remains firmly grounded in the notion
that arbitrators must be natural persons, as mentioned in Article 10 (1)(a).
Moreover, Article 1 defines an ‘arbitral tribunal’ as a sole arbitrator or a
panel of arbitrators, with no recognition of the possibility that a machine or
algorithm could fulfill this role. Further, Article 10 (2), which governs the
appointment of arbitrators, uses terminology such as ‘nationality,’ a concept
that can only apply to natural persons, thereby excluding AI systems. The
requirements for issuing arbitral awards also reinforce this interpretation:
Section 41 mandates that awards must be in writing, signed by the members of
the arbitral tribunal, and supported by reasons. These obligations presuppose
human reasoning and signature, which AI systems currently cannot provide under
UAE law.[9]
In the same instance, in Lebanon, the legal framework governing arbitration
remains ill-equipped to accommodate electronic or AI processes. Under domestic
arbitration rules, Lebanese Code of Civil Procedure (LCCP) (LCCP, arts.
762–808), arbitration clauses must be in writing and signed to be valid, which
excludes electronically concluded or AI-generated agreements from recognition.
Even arbitral awards require production in hard copy, further restricting
reliance on digital or AI-based procedures. While the LCCP adopts a more
flexible stance towards international arbitration (arts. 809–821) by not
expressly requiring written agreements, recognition and enforcement of
international awards in Lebanon still depends on submission of original or
certified copies of both the arbitration agreement and award. This insistence
on material documentation, coupled with a conservative business culture
favoring face-to-face dealings, creates significant obstacles to the
integration of AI technologies into arbitration practice.[10]
On the other hand, Qatar has taken notable institutional steps to modernize its
arbitration framework and integrate digital technologies that pave the way for
artificial intelligence in dispute resolution. A key development was the 2024
Memorandum of Understanding between the Qatar International Court and Dispute
Resolution Centre (QICDRC) and Jus Mundi, an AI-powered arbitration knowledge
platform. This cooperation enhances access to case law, arbitral awards, and
legal materials through AI-driven search and summarization.[11]
Complementing this, the Qatar International Centre for Conciliation and
Arbitration (QICCA) introduced its revised arbitration rules in 2024, which
came into effect in early 2025. These rules expressly provide for electronic
communication, digital filings, and the electronic signature of awards under
defined safeguards. The modernization of QICCA’s rules, combined with
electronic services such as online claim submissions and case management,
creates a legal infrastructure conducive to the deployment of AI tools for
tasks such as document review, predictive analytics, or drafting assistance.[12]
Furthermore, regarding
international laws and regulations, we refer to the New York Convention's
provisions on arbitrators. The phrase "arbitral award" is used in the
New York Convention under Article I (2)[13]
to refer to decisions made by permanent arbitral bodies as well as decisions
made by arbitrators selected to handle specific disputes. Due to this, we can
deduce that the New York Convention empowers both natural legal persons to
issue international arbitral awards that are recognized and enforceable.
Moreover, most
legislation, along with the New York Convention, requires the rendered arbitral
award to be issued in a written form and have the signature of the concerned
arbitrators. Article IV(1)(a) of the New York Convention mandates submission
of the ‘duly authenticated original award, or a duly certified copy thereof’,
signaling that signed awards are presumed essential for enforceability.[14]
Felipe Volio Soley has pointed out that although electronic signatures are now
common, e‑signatures were not envisioned in the Convention and may still
raise concerns about formality in some jurisdictions.[15]
Meanwhile, the LCIA Rules 2020 (Art. 26.2) explicitly permit electronic
signatures by arbitrators—provided parties or the tribunal do not object—but
this applies solely to human arbitrators.[16]
Finally, a recent commentary raises the question of whether an AI‑only
award, lacking any human authentication, would meet the enforceability and
public‑policy thresholds under the Convention.
On the other hand, this
clause is not included in the UNCITRAL Model Law. However, Article 11(1)[17] of
the Model Law provides that no person can be prohibited from serving as an
arbitrator due to his or her nationality. We can deduce that this provision
refers to human arbitrators.
Moreover, in a similar
context, nationality and the arbitrators' domicile are included in the ICC
Arbitration Rules[18] under
Article 13(1)[19],
(5)[20],
and concerning Article 6.1[21]
of the Arbitration Rules of the London Court of International Arbitration
(LCIA)[22],
there are specific guidelines on the nationality of the arbitrators that
consider dual nationality.
Yet, on the other hand,
we should note that despite the analysis made above, the New York Convention
states nothing regarding whether AI arbitrators can function as arbitrators and
replace human arbitrators. Some believe that the convention was approved in
1958, and it is evident that the idea of choosing AI arbitrators was not taken
into consideration at that time. Also, others believe that such an omission
might mean that the Convention does not contain any limitations on awards
issued by AI arbitrators and therefore there are no limitations on the validity
and enforcement of awards by AI arbitrators or even on the ability for AI
arbitrators to replace human arbitrators.[23]
Legal experts believe that this interpretation can be made similar to the
interpretation that was made in 2006 when UNCITRAL expanded the scope of Art.
II(2) of the Convention to include arbitration agreements reached via email or
other instantaneous modes of communication.[24]
Moreover, as for the
legal requirement relating to the legal personality of AI systems and tools, it
should be referred to and noted that although the European Parliament adopted
in 2017 the Resolution on Civil Law Rules on Robotics, these conversations are
still ongoing and have not yet been the subject of legislation, especially in
the field of arbitration.
In addition to that,
reference to the EU Artificial Intelligence Act, enacted in 2024 (the “AI
Act”), published on 21 April 2021, that set out regulations and rules on
artificial intelligence should be made. This AI Act followed the publication of
the European Strategy on Artificial Intelligence in 2018, and the publication
of the Guidelines for Trustworthy AI in 2019 by the High-Level Expert Group on
Artificial Intelligence, established by the European Commission to make sure AI
is created and deployed in a manner that is secure, transparent, upholds
ethical standards, encourages innovation, and boosts the economy.
This AI Act, besides
defining artificial intelligence and categorizing AI systems, has addressed in
Title III the high-risk AI systems. The AI Act has stated the types of
high-risk AI systems in Annex 3; some of these high-risk systems are the AI
systems used for law enforcement and for assisting judicial authorities in
researching and interpreting facts and the law, and in applying the law to a
concrete set of facts[25]
which the use of such systems in the legal field is subject to strict
requirements.
We can summarize that the
AI Act includes provisions related to the use of AI systems in law. and it
acknowledges that applying AI to the legal industry has the potential to
increase decision-making support, increase access to justice, and boost process
efficiency. It also recognizes that using AI in law may give rise to issues
with morality that need to be resolved and addressed.
The AI Act distinguishes
between different categories of AI systems, based on their potential risk to
individuals, society, and the environment. For example, AI systems that are
used to make decisions with legal effects, such as in court proceedings, are
likely to be classified as high-risk systems and subject to strict
requirements.
The AI Act mandates that
AI systems used in areas of the law be created and built in a way that
guarantees transparency, accountability, and fairness. This means that users of
AI systems must be informed about how the system works and the data used to train
it. Additionally, the AI Act requires that AI systems used in legal contexts
are subject to human oversight, meaning that there must always be a person
responsible for the final decision.
Overall, the AI Act
recognizes the potential benefits of AI in the legal field but emphasizes the
need for responsible use and adequate safeguards. The regulation aims to ensure
that AI systems used in law respect fundamental rights and ethical principles
and contribute to the promotion of human well-being.
Adding to the above, as
the scholars Marrow, Karol, and Kuyan point out, much of the difficulty lies in
whether current legal instruments even permit arbitrators who are not natural
persons. Their findings underscore the uncertainty of enforcing AI-issued
awards.[26]
Scholar Horton further warns that meaningful consent is at stake: even if
parties appear to consent to an AI arbitrator, the fairness of such consent may
be questioned under mandatory arbitration law.[27]
In my opinion, the current debate should focus less on whether AI could
theoretically serve as an arbitrator and more on whether existing arbitration
laws and international conventions are capable of accommodating such a
development. Until there is explicit recognition of non-human arbitrators, the
enforceability of AI-issued awards will remain uncertain. Thus, the issue is
not simply one of party autonomy, but also one of institutional legitimacy and
legal certainty.
Yet, besides the legal
normative approach, the question of the possibility of AI taking the role of
human arbitrators highlights the important role that public policy could have
in answering this ongoing debate.
1.2
The Social Context
An important factor in
international arbitration, across Inter-State, International Investment, and
International Commercial Arbitration, is public policy. It has been known as a
blanket term in international arbitration since it includes violations of due
process to violations of the fundamental applicable laws and regulations.[28]
The New York Convention and the UNCITRAL Model Law both consider the forum's
public policy. This is why there is no unified and universal definition for
public policy since its definition may change depending on the forum state. As
a result, it's crucial to consider the public policy of the state wherever
procedures are being challenged or where a request is being made for the
recognition or enforcement of an award. A state's legal, economic, and social
order, as well as the fundamental rights and liberties of its inhabitants, as
well as social norms and moral standards, may all be defended by certain
principles, which are referred to as public policy ideals. Empirical data from
the author's survey underscores that practitioners value enforceability and public
trust over full automation, reflecting why public policy exceptions may remain
robust in resisting AI-only awards.
However, when answering
the question of the possibility of AI replacing human arbitrators, several
issues regarding public policy can arise when arbitral awards are rendered and
issued by AI arbitrators. One of the issues is when the law of the forum requires
the arbitral award to be issued by a natural arbitrator. In this case, the
arbitral awards issued by AI arbitrators can be placed aside and rejected as a
result of the laws and public policy of the forum. The New York Convention has
reflected this position in Article V(2)(b)[29],
where it is stated clearly that the recognition and enforcement of arbitral
awards are rejected and refused if the competent authority finds that this
recognition and enforcement of the award is contrary to the applicable public
policy.[30]
Therefore, we can deduce that the interpretation of whether the breach of
public policy took place should be determined. These legal interpretations can
be tested against the survey results in Section 2, where the overwhelming
majority of practitioners indicated that AI is currently used to support,
rather than replace, human arbitrators. This alignment suggests that current
legal constraints reflect prevailing professional attitudes. This concern is
reinforced by the survey's finding that 93.1% of respondents believe AI tools
should support, not replace, arbitrators—indicating a strong alignment between
professional sentiment and public policy caution. Empirical data from the
author's survey underscores that practitioners value enforceability and public
trust over full automation, reflecting why public policy exceptions may remain
robust in resisting AI-only awards.
Another issue that could
arise is related to the impartiality and independence of AI arbitrators. This
issue can be raised since AI arbitrators are considered data machines that
function based on the collected data. Therefore, the data inserted and collected
by the AI arbitrator can be discriminatory, which would lead to the arbitral
award issued by them being discriminatory, causing impartiality issues. It
should be stated that although impartiality and discriminatory problems can
also occur in human arbitrators’ situations, having AI machine arbitrators
functioning on the data collected and without emotional intelligence may
increase the possibility of discrimination.
In addition to what has
been stated above, another issue could be related to the reasoning of arbitral
awards, since the lack of reason in the arbitral awards issued by AI
arbitrators may also be considered a violation of public policy. According to
Article 31 (2) of the UNCITRAL Law Model[31],
unless the parties have agreed that there are no grounds to express, the
arbitral award must include the reasons upon which it was given. Several pieces
of legislation in multiple jurisdictions require any decision or arbitral award
issued to include reasons as a requirement of justice to help parties
understand why and how this decision was taken. As a result, in these nations,
if AI arbitrators make arbitral rulings without sufficient justification, it is
viewed as a breach of public policy and will result in the arbitral award being
invalidated since the parties' entitlement to a just and reasoned award was not
honored. Finally, according to several experts’ opinions, the idea of AI
arbitrators replacing human arbitrators is itself a violation of public policy
due to the fear of the unknown.[32]
In other words, several jurisdictions and international frameworks require
arbitral awards to include clear reasons; this requirement ensures fairness,
transparency, and the parties’ understanding of the decision. Courts may view
an AI arbitrator's unexplained “black box” decision-making as a breach of these
fundamental values, particularly if it compromises impartiality or justice.
Indeed, scholars have highlighted that an award rendered without transparent
reasoning conflicts with parties’ rights to a fair trial. For instance, under the
Korean Arbitration Act, courts may set aside awards that violate public policy
or “good morals”.[33]
This concern is reinforced by the survey's finding that 93.1% of respondents
believe AI tools should support, not replace, arbitrators, indicating a strong
alignment between professional sentiment and public policy caution.
To conclude, assessing
whether AI-issued arbitral awards violate public policy requires a nuanced
analysis of how such awards align with a state's legal framework and societal
values. Ultimately, the way national courts respond to the growing integration
of technology in arbitration will shape the application of public policy
exceptions.
2.
Data and Information
In the current era of
technological advancement that has been accelerated by the COVID-19 pandemic
(which was classified by the World Health Organization as a global pandemic),
the legal field, especially international arbitration, has been facing a
technological transformation in every aspect. The core of such transformation
is the increased use of AI technological tools. AI technological tools have progressed
to a level that these tools are being used to predict the result of disputes,
analyze arbitration clauses, and even draft arbitration clauses, in addition to
several other tasks performed, that raised the question of the possibility of
AI arbitrators taking the role of human arbitrators. This ongoing debate and
concern are because some countries have already applied the use of AI judges
and arbitrators in the decision-making process. For example, China has
developed digital courts that are handled by AI judges.[34]
Moreover, this ongoing
debate is also a result of the arbitration community’s desire to apply
technological tools that increase the efficiency and modernization of
international arbitration. The survey and study conducted by the School of International Arbitration at Queen Mary
University of London, in partnership with White & Case LLP, in 2021
titled “2021 International Arbitration Survey: Adapting Arbitration to a
Changing World” has examined how international arbitration has managed to adapt
to the ongoing changes. This survey was considered to be the survey with the largest
number of respondents, with 1218 questionnaire responses received and 198
interviews conducted.[35]
The respondents were from diverse backgrounds, including in-house counsel from
both public and private sectors, arbitrators, private practitioners,
representatives of arbitral institutions and trade associations, academics,
experts, and third-party funders.[36]
One of the chapters of this survey, titled “Use of Technology: The Virtual
Reality,” investigates the usage of technological tools in 2021 in comparison
to the results that were gathered in 2018. This study was conducted by
inquiring about of often certain technological tools are used, one of which is
AI. As seen from the survey, when we compare the results of 2021 with the
results recorded in 2018, we can deduce that the use of technological tools
such as videoconferencing, hearing room technologies, and cloud-based storage
remained the same without any increase or decrease in their percentage of
usage. However, it shows a significant increase in the use of AI tools.
According to the result
chart, 35% of the respondent group stated that they have 'never' used AI, while
24% stated that they have used AI rarely. Only 15% have used AI 'frequently' or
'always'.[37]
It should be stated that,
according to this study and the interviews with respondents, it has been shown
that the increased use of AI tools refers to the tools used for data analysis
and document review. It has been discussed that AI technologies are considered
to be expensive and mostly unaffordable. Even if parties were able to afford
their price, they are not always convinced that the value produced by these AI
technologies will outweigh the hefty expenditures they involve. Moreover,
interviews also found that there is a persistent fear of new technology and a
general lack of familiarity with it, whether from a practical or legal aspect.
All of the interviewees indicated some skepticism about the possible
application of AI tools and algorithms for predictive justice. They brought up
moral questions and questioned how much such instruments should or may obstruct
the adjudicative process. The overwhelming majority of
participants believed that human arbitrators and counsels cannot be replaced by
AI arbitrators.
In addition to the
above-mentioned date and as part of the author’s research into the field of
arbitration and the use of AI tools, the author conducted a survey tackling a
small group of professional and experienced arbitration practitioners from
diverse nationalities and backgrounds. Although the author's survey did not
gather as many respondents as the School of International Arbitration's survey
conducted in partnership with White & Case LLP at Queen Mary University of
London, the results were similar and supported those obtained by the latter
survey. The purpose of the survey conducted by the author was to gain insights
through the experiences and perspectives of the respondents on whether AI tools
are used in a manner that will allow them to replace human arbitrators or if AI
tools are just used to provide support to arbitrators and the arbitration
field. Through this small survey, the author hopes to contribute to the broader
conversation on arbitration and the use of AI tools and provide valuable insights
for those involved in this field. Noting that the survey was conducted on a
small but diverse group of arbitration practitioners with various backgrounds
and nationalities. This original survey, designed and implemented by the
author, serves as an empirical contribution to the debate.
According to Figure (1)
below, the first question posed to the respondents was about the frequency of
their use of AI tools in the arbitration field. The results indicate that most
respondents use AI tools in their profession, but at different levels and
frequencies. Specifically, 13.80% of the respondents use AI tools frequently,
while 20.70% use them occasionally. Additionally, 27.60% use AI tools somewhat
rarely, while another 27.60% reported using them very rarely. Finally, 10.30%
of
respondents
declared that they do not use AI tools in their arbitration field.
|
|
 |
We
can deduce that a small percentage of respondents do not use AI tools in their
arbitration field compared to the other percentages. This finding supports a
study conducted by the School of International Arbitration at Queen Mary
University of London, in partnership with White & Case LLP in 2021, which
investigated the usage of technological tools in arbitration. Both studies
indicate an increase in the use of AI tools in the arbitration field.
In
addition, the second question shown in Figure 2 below was about the types of AI
tools arbitrators use in their arbitration field. The results indicate that
62.10% of respondents use AI tools that support legal research and analysis,
while 20.70% use AI tools for document review and analysis. Furthermore, only
6.90% of respondents use AI tools that provide predictive analytics software
and e-discovery tools. Surprisingly, none of the respondents reported using AI
tools that provide decision support systems. Therefore, based on the results in
Figures 1 and 2, we can deduce that most arbitrators use AI tools at different
frequencies; however, they tend to limit their use to tools that support legal
research, document review, and predictive analysis, without the use of decision-making
support systems which support the results of the study conducted by the School
of International Arbitration at Queen Mary University of London and in
partnership with White & Case LLP in 2021
|
|
 |
(mentioned
above). Moreover, Figure 3 below shows whether respondents believe that AI
tools are used in a manner that allows them
to replace human arbitrators or if they are used in a way that provides support
to arbitrators. The results indicate that 93.10% of respondents believe that AI
tools are used in a manner that provides support to human arbitrators, while
only 6.90% of respondents believe that AI tools are used in a manner that would
replace human arbitrators. Also, the results in Figure 4 below, along with the
respondents' short answers on their beliefs regarding the aspects in which AI
will assist or replace human arbitrators, show that 51.70% of the respondents
believe that AI will partially replace human arbitrators. They believe that AI
will replace human arbitrators in aspects related to data collection, analysis
software, summarizing evidence, legal research, and document review, since the
use of AI tools in these aspects will assist arbitrators and help them save
time and solve disputes more efficiently. On the other hand, 44.80% of the
respondents believe that AI will not replace human arbitrators. According to
their short answers on the aspects in which AI will assist or replace human
arbitrators, some stated that although AI could assist in having a quick and
efficient arbitration process, it will not replace human arbitrators.
It should be noted that
only 3.40% of the respondents believe that AI would completely replace the role
of human arbitrators. Therefore, we can deduce that the percentage of
respondents who believe that AI would completely replace human arbitrators is
very low compared to the percentage of respondents who believe that AI would
not or would partially replace human arbitrators by providing support for a
faster and more efficient process. Moreover, these results support and are in
line with the findings of the study conducted by the School of International
Arbitration at Queen Mary University of London and in partnership with White
& Case LLP in 2021, which showed that the majority of participants believed
that human arbitrators and counsels cannot be replaced by AI arbitrators.
|
|
 |
Moreover,
Figure 5 below shows the results regarding whether AI arbitrators could provide
faster and more accurate decisions than human arbitrators. The results indicate
that 58.60% of respondents believe that AI arbitrators could not provide faster
and more accurate decisions compared to the 6.90% of respondents who believe
that they could. It is also significant to mention
that 34.50% of respondents answered "maybe" to this question.
Last but not least, Figure 6 below
addresses the question of how the role of human arbitrators is changing with
the increased use of AI, on a scale of 1 to 5. The figure shows that 31% of
respondents believe the role is slightly changing, while only 6.90% believe it
is very
changing.
|
|
 |
We
can conclude from the data, information, and analysis provided above that there
has been a significant increase in the use of AI tools by arbitration
practitioners in the arbitration field. However, this increased use of AI tools
focuses on the AI tools that assist arbitrators in increasing efficiency and
saving time, rather than replacing them from a decision-making perspective.
Moreover, the results have shown that it is believed that the increased use of
AI tools is limited to supporting arbitrators rather than replacing them
completely, as the belief that faster and more accurate decisions remain a
significant characteristic of human arbitrators, not an AI arbitrator. Notably,
the empirical findings both support and contextualize the legal analysis above:
while laws and conventions implicitly require human arbitrators, practitioners'
overwhelming preference for AI as a supportive tool rather than a replacement
suggests that legal norms are not merely restrictive but are consistent with
professional consensus.
- Discussion
Scholars Marrow, Karol,
and Kuyan (2020) left the question of 'Are we there yet?' open. This study’s
findings suggest the answer remains negative under current conditions, though
future reforms may shift the balance.[38]
Scholar Horton, however, urges caution, warning that premature or forced
adoption may compromise fairness and erode confidence in arbitration as a
trusted mechanism.[39]
In my opinion, AI tools
have advanced to the stage where they can perform many of the tasks that an
arbitrator can. The increasing use of AI tools in arbitration is significant.
Time has shown that no profession is immune to having AI tools take control of
some aspects of its work, including the role of an arbitrator. By processing
materials in a way that is similar to human arbitrators, AI poses a challenge
to the international arbitration field. This technological development and
rapid adoption of new tools have led to this article addressing the ongoing
debate over whether AI arbitrators could replace human arbitrators from legal
and social perspectives. In other words, the article examines whether AI tools
are used to replace human arbitrators or merely to support them and the
arbitration field.
From a legal perspective,
this article has analyzed and studied several pieces of legislation,
conventions, and instruments used by arbitration organizations to determine
whether AI can replace human arbitrators according to current applicable
regulations. It has been realized that the legal framework of arbitration is
designed to have human arbitrators as decision-makers, whether this is stated
clearly in the regulations or indirectly deduced from the requirements for
arbitral awards, arbitrators, and the arbitral process in general. Noting that,
in my opinion, only time can confirm if AI tools can replace human arbitrators.
Moreover, it should be
noted that implementing the use of AI tools and AI arbitrators without suitable
regulations on their use and application in the arbitration field could impact
arbitration's reputation as a preferred dispute resolution mechanism. Therefore,
the EU's proposal for the Artificial Intelligence Act (“the AI Act”), published
on 21 April 2021, is considered a significant step toward modernizing the
applicable regulations. The AI Act distinguishes between different categories
of AI systems based on their potential risk to individuals, society, and the
environment. It mandates that AI systems used in areas of the law be created
and built in a way that guarantees transparency, accountability, and fairness.
This means that users of AI systems must be informed about how the system works
and the data used to train it. Additionally, the AI Act requires that AI
systems used in legal contexts are subject to human oversight, meaning that
there must always be a person responsible for the final decision.
In other words, the AI
Act recognizes the potential benefits of AI in the legal field but emphasizes
the need for responsible use and adequate safeguards. The regulation aims to
ensure that AI systems used in law respect fundamental rights and ethical principles
and contribute to the promotion of human well-being.
From a social
perspective, public policy plays a significant role in the use of AI in
arbitration and the possibility of AI arbitrators replacing human arbitrators.
When answering the question of whether AI can replace human arbitrators,
several issues regarding public policy can arise when arbitral awards are
rendered and issued by AI arbitrators, especially if they violate state law and
societal norms. As mentioned earlier, how courts, laws, regulations, and
society respond to technology will determine how public policy grounds are and
will be applied. This concern is reinforced by the survey's finding that 93.1%
of respondents believe AI tools should support, not replace arbitrators, indicating
a strong alignment between professional sentiment and public policy caution.
Furthermore, the
statistical studies, results, and data mentioned in Part III help in having an
initial answer to the ongoing debate. The results of studies, whether conducted
by the author or by the School of International Arbitration at Queen Mary University
of London and in partnership with White & Case LLP, demonstrate a
significant increase in the use of AI tools by arbitration practitioners in the
arbitration field. However, this increase despite its frequency is focused on
tools that assist arbitrators to increase efficiency and save time, rather than
replacing them from a decision-making perspective. Additionally, the results
have shown that most respondents agree that AI tools are only used in a manner
that supports arbitrators and would not replace human arbitrators. In my
opinion, this indicates a cautious and pragmatic approach, one that
acknowledges the benefits of technological innovation while safeguarding the
essential role of human judgment, fairness, and legitimacy in arbitration.
Conclusion
This article has examined
the potential for AI arbitrators to replace human arbitrators, considering both
legal and social perspectives. The analysis suggests that, within the current
legal framework, AI arbitrators are unlikely to fully replace human decision-makers.
Existing arbitration laws emphasize the requirement of human judgment, consent,
and due process, making legislative reform a prerequisite for AI to assume such
a role. Socially, too, parties continue to prefer the assurance of human reasoning,
accountability, and empathy in adjudication.
At present, AI tools are
most effectively deployed as support mechanisms assisting with document review,
translation, case management, and predictive analytics, thereby enhancing
efficiency while leaving final decision-making to human arbitrators. Looking
forward, however, incremental reforms could progressively broaden the role of
AI. Legislators and arbitral institutions should consider:
·
Regulatory innovation: adopting clear
guidelines that permit AI to assist in defined procedural or evidentiary tasks
while safeguarding party autonomy and due process.
·
Transparency and disclosure
standards: requiring parties to be informed of how AI is used during
proceedings, thus maintaining fairness and confidence.
·
Ethical oversight frameworks:
creating specialized committees or adapting institutional rules to monitor the
risks of bias, data security, and accountability in AI-assisted arbitration.
·
Capacity building: training
arbitrators and practitioners in the responsible use of AI tools to ensure
equal access and reduce technological asymmetry between parties.
From my perspective, AI
arbitrators may, in the medium term, take on routine or procedural functions,
but full substitution of human arbitrators remains unlikely in the foreseeable
future, given legislative, cultural, and public policy constraints.
Nevertheless, with thoughtful reforms, AI could evolve from a mere support tool
into a trusted partner in arbitration, enhancing efficiency without eroding the
human values that underpin justice. Yet, the key question remains: will the
future of arbitration witness AI as a co-pilot to human judgment, or as a
full-fledged arbitrator in its own right?
Bibliography
I.
Books
& Book Chapters
II.
Articles
& Online Journals
III.
Websites
& Blog Posts
IV.
Statutes
V.
Conventions & Rules
VI.
Master’s
Thesis
Acknowledgements:
International 
المجلة
الدولية
للقانون