Machine Arbitration and Machine Arbitrators

By Jack Wright Nelson (King & Wood Mallesons)

Artificial Intelligence (“AI”) will revolutionise legal practice. Over the next 20 years, the technologies with which we currently practice law will themselves begin to practice law.

These changes are starting now. Last month, an American law firm began using an AI researcher that can conduct legal research faster and cheaper than a human. Some commentators predict the mass redundancy of junior lawyers all together.

Arbitration is not immune to this technological advancement. Imagine an AI lawyer that is capable of understanding argument, ascertaining facts, and determining the applicable law. There is one particular role that would treasure the neutrality and independence that such intelligent technology could provide – that of the arbitrator.

This post explores whether our current arbitral framework could allow such an AI lawyer, a “machine arbitrator”, to resolve disputes between parties who have consented to this “machine arbitration”.

Would a machine arbitration agreement be enforced?

In our hypothetical scenario, a contractual dispute has arisen between Alpha Limited (“Alpha”) and Omega Corporation (“Omega”). Both are commercial entities, carrying on business in Hong Kong.

Fortunately, the relevant contract between Alpha and Omega has a dispute resolution clause, Clause 22.

“Clause 22: Settlement of Disputes

22.1 Any dispute, controversy or claim arising out of or relating to
this contract, or the breach, termination or invalidity thereof,
shall be settled by arbitration in accordance with the UNCITRAL
Arbitration Rules.”

So far, so standard. But Alpha and Omega have amended the standard clause, as follows:

“22.2 The number of arbitrators shall be one.

22.3 The sole arbitrator shall be version 3.2 of the machine arbitrator program “DecisionMakerPlus,” as released by Dyno Corporation on 31 August 2022.

22.4 All references to the “arbitrator” in the UNCITRAL Arbitration Rules shall be considered as references to the machine arbitrator program stated in Clause 22.3 of this contract.

22.5 The place of arbitration shall be Hong Kong.

22.6 The language to be used in the arbitral proceedings shall be English.”

When negotiation fails to resolve the contractual dispute, Omega commences a court action against Alpha in Hong Kong. Can Alpha rely on Clause 22 to stay these court proceedings?

Hong Kong’s Arbitration Ordinance (Cap. 609) (“Arbitration Ordinance”) incorporates the UNCITRAL Model Law on International Commercial Arbitration (“Model Law”). Section 20 of the Arbitration Ordinance reproduces Article 8(1) of the Model Law, providing that:

“[a] court before which an action is brought in a matter which is the subject of an arbitration agreement shall … refer the parties to arbitration unless it finds that the agreement is null and void, inoperative or incapable of being performed.” (emphasis added)

Applying Article 8(1) to Clause 22 gives rise to two questions. First, whether Clause 22 is an “arbitration agreement” for the purposes of the Model Law. Second, whether Clause 22 is “null and void, inoperative or incapable of being performed”.

The answer to question one seems to be yes. Clause 22 clearly states that the parties have waived their right to have their disputes resolved in court. In doing so, it grants jurisdiction over their dispute to an arbitrator. Accordingly, Clause 22 meets the definition of arbitration agreement stated in the Model Law:

“An agreement by the parties to submit to arbitration all or certain disputes
which have arisen or which may arise between them in respect of a defined
legal relationship, whether contractual or not”.

The answer to question two seems to be no. “Null and void” and “inoperative” refer to situations where the validity of the arbitration agreement has been impugned. In our hypothetical, no such issues arise. The contract between Alpha and Omega is valid and has not been affected by duress or any other vitiating factors.

Whether Clause 22 is “incapable of being performed” is a less straightforward question. Assuming (as we are) that DecisionMakerPlus exists, then it appears that Clause 22 is capable of being performed.

However, the biggest challenge would be demonstrating this fact in court, as judges may not be familiar with machine arbitrators such as DecisionMakerPlus. Fortunately, this issue is not intractable. It would likely arise in the early days of machine arbitration, before familiarity with the relevant technology spreads throughout the legal profession.

Would a machine arbitrator’s award be set aside?

Imagine now that the machine arbitration has taken place and DecisionMakerPlus has rendered a written, reasoned and signed award in Alpha’s favour (“the Award”).

The Award complies in every aspect with Section IV of the UNCITRAL Arbitration Rules. But Omega refuses to honour the Award, and applies to the Hong Kong court to have the Award set aside. Is Omega likely to succeed?

Section 81 of the Arbitration Ordinance reproduces Article 34 of the Model Law. This Article provides the limited grounds on which arbitral awards may be set aside. Two of these grounds, Articles 34(1)(a)(iv) and 34(2)(b)(ii) may be opposable to the Award.

Article 34(1)(a)(iv) provides that an arbitral award may be set aside if:

“the composition of the arbitral tribunal or the arbitral procedure was not in
accordance with the agreement of the parties, unless such agreement was in conflict with a provision of this Law from which the parties cannot derogate, or, failing such agreement, was not in accordance with this Law.”
(emphasis added)

In our hypothetical situation, both the composition of the arbitral tribunal and the arbitral procedure was clearly in accordance with the relevant arbitration agreement, namely Clause 22. But did Clause 22 conflict with a mandatory provision of Hong Kong law?

The answer seems to be no. Nowhere does the Arbitration Ordinance state that arbitrators must be human or that the arbitral procedure must include a human element.

The Arbitration Ordinance does contain numerous assumptions that arbitrators are human, most notably by adopting the Model Law’s consistent use of the pronouns “he” and “his”. However, Section 7(1) of Hong Kong’s Interpretation and General Clauses Ordinance (Cap. 1) provides that “[w]ords and expressions importing the masculine gender include the feminine and neuter genders.” In English, a machine arbitrator would most likely take the neuter pronoun (“it”). Accordingly, it seems that the use of masculine pronouns does not (at least in Hong Kong) imply a requirement that the arbitral tribunal comprise humans.

The second ground for setting aside that is potentially opposable to the Award is Article 34(1)(b)(ii) of the Model Law. This Article provides that an arbitral award may be set aside if:

“the award is in conflict with the public policy of this State.”

Hong Kong courts typically adopt a narrow approach to public policy, only setting aside arbitral awards when fundamental notions of morality and justice have been compromised.

Whether Hong Kong courts would consider that the Award did compromise these fundamental notions is, of course, an untested and open question. However, it is hard to deny that a machine arbitration would give rise to public policy concerns. Further, it is highly likely that if Alpha sought enforcement outside of Hong Kong, then some national courts would refuse enforcement on public policy grounds.

These concerns would likely impede the widespread adoption of machine arbitration. However, if the commercial users of arbitration consistently elected for machine arbitration to resolve their commercial disputes, then public policy concerns would eventually recede in the face of this consent.

Indeed, in many jurisdictions arbitration was itself considered to be contrary to public policy through its displacement of national courts. Today’s judicial recognition of the legitimacy of arbitration shows that with time and experience comes familiarity, and ultimately acceptance.


The advent of AI will revolutionise legal practice. This post has offered one perspective on the potential impact for those working in the arbitration arena, arguing that there is no explicit bar to machine arbitration or machine arbitrators in one Model Law jurisdiction. Public policy concerns are likely to arise, but these are posited to diminish over time.

The exponential nature of technological advancement necessitates prompt action to address the legal, ethical and practical challenges raised by machine arbitration. The development of specialised arbitral rules and frameworks to govern and administer machine arbitration will be essential should this technology become reality.

Perhaps these rules and frameworks should be drawn up today. After all, the Permanent Court of Arbitration published their Optional Rules for Arbitration of Disputes Relating to Outer Space Activities in 2011, well in advance of widespread commercial spaceflight. Similar foresight in respect of machine arbitration may minimise the hazards that tend to accompany consequential change.