Revue - Les Cahiers de l'Arbitrage N°2 - 2024 - 53

DOCTRINES - DÉBATS
(i) Pro-forma Decisions
The two key areas that counsel at institutions are engaged in are (1) compliance
with the rule requirements for the notice of arbitration (NOA) and/or the request
for arbitration (RFA) and (2) existence of a prima facie valid arbitration agreement.
Both tasks can be performed by a specific type of AI, i.e., " supervised machine
learning " . This essentially means that a human being can train the algorithm by
specifying a set of desired outcomes within a given range of inputs. 45
In other
words, the AI could be programmed to do the work that counsel at an institution
does - compare the NOA/RFA against the requirements in the rules, and only
where there is a 100% match, confirm compliance with the same. Where there is
less than a 100% match, the AI could make a recommendation for counsel at the
institution to review it.
Similarly, the AI could also be " instructed " on what kind of language/key words
an arbitration agreement must possess in order to be considered valid. If there is
a 100% match with the set of key words given, the arbitration agreement should
be considered prima facie valid; similarly, for anything below 100%, the AI could
make a recommendation for counsel to review it. In this way, the institution would
save time and cost, but also ensure that due process is not jeopardised.
(ii) Complex Decisions
In recent years, arbitral institutions have introduced provisions in the rules to
deal with multi-party disputes and the consolidation of arbitrations, joinder of third
parties and other multi contract/multi party scenarios. Such provisions address the
gap between the commencement of an arbitration and the constitution of a tribunal
by empowering the institution/court to make the necessary decisions for the case to
go forward. By setting certain criteria for such findings, institutions standardise the
practice to meet the intended expectations of parties. However, this exercise can
take months to resolve, especially given that there has been an increasing number
of applications requesting institutional intervention. 46
While many of these scenarios are fact-specific, and procedural issues arising
from complex disputes may require more human intervention in the earlier stages
of the case, there are many circumstances that could be dealt with by AI.
The HKIAC offers an " insight " into its procedural decision-making, through
the HKIAC Case Digest platform. This also includes access to the " analyses of
HKIAC's standing committees " , decisions on different procedural issues and " case
abstracts " . 47
Such data could train AI to understand the parameters for such complex
45. P. Flach, Machine Learning: The Art and Science of Algorithms that Make Sense of Data,
Cambridge University Press, 2012. See also A. Gadkari and A. Malhotra, " Weaving Arbitral Awards
from the Mills of Artificial Intelligence " , International Journal of Law and Policy Review, 2012,
Vol. 11(2); and M. Scherer, " Artificial Intelligence and Legal Decision-Making: The Wide Open? " ,
in M. Scherer (ed), Journal of International Arbitration, (Kluwer Law International 2019), Vol. 36(5),
pp. 539-574.
46. Institutions will generally keep records of such decisions. HKIAC has in fact published such
decisions, making them transparent, which helps inform future applicants.
47. See the HKIAC Case Digest, available at https://hkiac.org/Case-Digest, last accessed on
15 June 2024.
Les Cahiers de l'Arbitrage 2024‑2
419

https://www.hkiac.org/Case-Digest

Revue - Les Cahiers de l'Arbitrage N°2 - 2024

Table des matières de la publication Revue - Les Cahiers de l'Arbitrage N°2 - 2024