The way we use artificial intelligence has changed a lot since its first appearance, handling boring jobs such as reviewing papers or agreements. We see that a significant development is artificial intelligence functioning as a standalone arbiter, empowered to deliver conclusive rulings for worldwide conflicts. This advancement is quite transformative, for it challenges the old notion of primary human responsibility in adjudication. Ultimately, this transition is motivated by the expanding necessity for economical and rapid tactics for handling disputes in international trade. However, this transition casts severe doubts on whether the accuracy and fairness of machine-driven decisions can be reconciled with the fundamental principles of arbitral justice^[1]. 

This article attempts to find a middle ground between technological advancements and concerns surrounding AI in arbitration the law by delving into three connected questions. First, do AI-generated awards meet the enforceability criteria set out in the New York Convention Second, what are the specific technical architectures like rule-based systems or machine learning models that can reproduce the sophisticated reasoning of an arbitrator? Third, in what manner can AI systems preserve transparency and fairness while reducing risks such as algorithmic bias? To address these questions, the article suggests an HOF. An HOF is a governance model that re-distributes decision-making powers between AI and human arbitrators according to the complexity and level of stakes of the disputes. This framework is intended to maximise efficiency when using AI while restricting its use in matters where human discretion is critical to ensure justice in arbitration. 

In terms of methods, this article takes an interdisciplinary approach that merges a doctrinal legal analysis of the New York Convention and the

United Nations Comm’n on Int’l Trade Law, Model Law on International Commercial Arbitration [“UNCITRAL Model Law”] with AI systems such as Brazil’s JIBOIA platform that automates the resolution of small claims. Ethical dimensions from digital ethics literature, including transparency and accountability, further shape the analysis. This scope of research is sparse but is addressed in parts by the European Commission’s 2021 study on the use of AI in commercial arbitration, which is the first case study on AI within the European Union [“EU”].

                             II.      Legal Foundation of AI Arbitrators 

Certainly, using artificial intelligence as arbitrators in international disputes creates problems under the New York Convention^[2]. Truly, this significant agreement, with over 170 nations as members, outlines the terms for accepting and implementing arbitration decisions in other countries. Drafted before digital adjudication existed, it presents both opportunities and challenges for AI-produced awards.

A. Legal Foundation of AI Arbitrators

One of the focal points of the New York Convention is Article II which highlights that for an award to be considered valid, it should be ‘in writing’ and ‘signed’ by the arbitrators. Although the Convention does not form technology-favourable wording, its use with AI systems bears some issues. For example, does a signature done through a digital ID generated by an algorithm fulfil the predefined ‘signature’ requirement of a human arbitrator? Countries such as Singapore and the EU have adopted laws on electronic signatures where the use of cryptographic authentication is accepted as evidence which provides a means of meeting the requirement.

For instance, Singapore’s Electronic Transactions Act 2021^[3] sets out the conditions in the law where signatures created through ‘automated electronic systems’ are accepted if they possess integrity and authenticity.

European Union [“EU”], like other jurisdictions, makes a provision for the acceptance of qualified electronic signatures which include AI executed signatures to be equivalent to handwritten signatures. EU does this through the electronic Identification, Authentication and Trust Services Regulation [“eIDAS”]^[4], However, jurisdictions with strict formalistic requirements like India and Argentina still mandate manual signatures for arbitral awards which leads to inconsistency in enforcement. This highlighted difference emphasises the importance of a more liberal interpretation of Article II of the New York Convention that considers the use of AI systems.  

Article V sets standards for enforcement restrictions based on due process and public policy. AI’s systematic workings threaten these principles. Algorithms unable to provide reasoning –’black box’ systems may fall below transparency thresholds expected in arbitration, risking invalidation under Article V(2)(b).^[5] To mitigate such risks, AI systems will need to be designed with a degree of explainability. Therefore, we need devices that have the ability to give justifications and produce valid rationale for the decisions rendered, like Local Interpretable Model-agnostic Explanations

[“LIME”]

B. Legal Personhood and Authority of AI Arbitrators     

Most legal systems do not recognise AI as having a legal personality, which makes its acceptance as an arbitrator difficult.^[6]

There are almost always gaps in scholarly argumentation, and such gaps are some of the most important parts where action can be taken. German professor Gunther Teubner formulates the idea of “digital legal persons”^[7], arguing that AI might achieve quasi-personhood if it is placed in an institutional environment that allocates rights and obligations. For example, there could be arbitration rules with liability insurances. Changes in legislations have started to provide alternatives.

In particular, UAE’s 2021 Arbitration Law^[8] states that ‘any natural or legal person’ may act as an arbitrator, which would allow for AI to act as a legal person. This is along the lines of Japan’s 2022 revisions to their Arbitration Act that allow for ‘non-human entities recognised by law,’ to serve as arbitrators with party approval. 

Conversely, France’s Article 1451 of the French Code of Civil Procedure^[9], requires arbitrators to be ‘physical persons,’ which suggests the necessity of harmonisation. The absence of an international agreement on the legality of AI invokes the problem of ‘enforcement havens’ where AI awards are accepted in some countries but not in others. As a result, everything that New York Convention seeks to unify will have the reverse effect.

C. Procedural Due Process in AI – Driven Arbitration 

The New York Convention and UNCITRAL Model Law emphasize proceeding fairness, opportunity to be heard, and proper jurisdiction. Incorporating these into AI systems ensures award enforceability. 

Key challenges include:

Transparency: The very basis of decisions reached through AI should be clear to the parties to the proceedings. Methods such as explainable AI [“XAI”]^[10] and post hoc rationalisation tools may provide some insight into the operations of algorithms as a prerequisite for due process. For example, the EU draft Artificial Intelligence Act, 2024^[11], seeks to impose ‘transparency logs’ on high-risk AI systems, which would compel arbitral algorithms to render their decision in procedures that are easy to understand.

Ethics of Responsibility: AI using historical data for its training has the potential to reproduce certain ingrained biases, which can harm fairness. Auditing for discrimination in algorithms regularly and using training datasets from different backgrounds is important for fairness compliance. The 2023 International Chamber of Commerce [“ICC”] Arbitration Commission Report^[12], suggested performing external audits on AI arbitration through the Organisation for Economic Co-operation and Development [“OECD”] Fairness Indicators to address the issue of discrimination.

Challenge Mechanism Procedures: There are no existing remedies for disputing the awards given by AI. The inclusion of human reviewers in the AI arbitration process, as suggested in the HOF, can solve some of these issues. The HOF’s approach where parties can raise disputes to human arbitrators from an AI source is like the appeal system of modern arbitration in which one seeks equity, even if not guaranteed, in the process.^[13]

• Party Autonomy and Consent 

The principle of party autonomy, which is critical for arbitration, seeks consent from parties in respect of AI arbiters. Article II (1) of the New York Convention is clear that there must be mutual consent for arbitration, and this is why arbitration agreements have to define whether a party consents to the proceedings being conducted by an AI. Institutions like the ICC could develop model clauses to try to reduce the vagueness surrounding such agreements. For instance, the 2022 London Court of International Arbitration [“LCIA”] Draft Clause for AI Arbitration^[14], which is the first known attempt to address arbitral clauses by including opt-in clauses that allow contracts to be honoured, to identify AI arbiters for preliminary procedural issues like document inspection, but not for the ultimate award. Still, the difference in the technological capabilities of the parties, in this case a multinational corporation and Small and Medium Enterprises [“SMEs”], raises the issue of whether there is informed consent and protective measures such as adequate disclosure of the shortcomings of AI is needed.

• Towards a Hybrid Legal –Technical Framework 

The HOF designs mechanisms enabling AI arbitration while addressing legal threats. In high-impact or ambiguous disputes, human arbitrators could review AI-generated awards ensuring compliance with the Convention’s procedural and substantive requirements. For example, in an inter-governmental construction dispute, an AI system might generate preliminary awards based on contracts and precedents while human arbitrators verify enforceability of provisions like anti-bribery clauses.

The HOF suggests developing worldwide certification protocols for AI arbitrators in an effort to address jurisdictional fragmentation. AI systems certified as complying with the convention would be subject to robust accountability as to their bias, accuracy, and transparency consistent with ISO/IEC 24028:2020, and would be given a ‘seal of enforceability’ claimable under the Convention. International organizations, such as UNCITRAL, could carry out certification to guarantee worldwide uniformity.

                        III.      Technical Feasibility of AI Arbitration 

The doing of AI as an arbitrator relies on the progress of machine learning, Natural Language Processing [NLP], and computerized decision making. While AI systems are now capable of performing a great deal of legal functions, customizing these systems to international arbitration is more difficult because they will have to deal with the interpretation of vague contractual clauses, assessments of a witness, and levels of conflicts of legal order. The input of AI to the legal domain involves two primary technologies; these are, rule-based systems and machine learning models. Automated systems that follow specific legal rules and logical systems are best suited for simple disputes with unambiguous terms such as algorithms for debt recovery that automatically calculate a penalty based on a set of conditions of payment.

One such modern case resolution tool is the online dispute resolution [“ODR”] platform Rechtwijzer from the Netherlands. Rechtwijzer facilitated the resolution of labour disputes, landlord–tenant conflicts, consumer complaints, and small business matters between 2015 and 2019, achieving an 80 per cent settlement rate. Settlements were reached through structured questionnaires and guided decision trees that directed parties toward mutually agreeable outcomes. However, such systems struggle with complex, high-stakes cases involving ambiguous contractual clauses or cross-jurisdictional regulatory conflicts.^[15]

France’s Prédictice is a perfect example of how a predictive justice tool uses the past to predict future decisions based on tribunal data and other variables. In Deep Learning and NLP models, the large database of awards, treaties, and case laws serve as a big dataset from which trends and future events can be predicted. In arbitration, machine learning would do the heavy lifting of cross-referencing witness statements against pre-existing legal and factual precedents and picking out contradictions and gaps in evidence provided by opposing witnesses.

As an illustration, a 2022 research segment conducted by the LCIA piloted an AI tool capable of processing more than 10,000 pages of evidentiary documents concerning a construction dispute, thus enabling a 60 per cent reduction in the document review phase.^[16] Nevertheless, the ‘black box’ nature of machine learning, where decisions are made without any rationale presented, can still compromise procedural justice.

Brazil’s JIBOIA platform adjudicates small commercial cases under $10,000 using NLP to examine claims, contracts, and documents, rendering awards within 72 hours. A 2022 evaluation found 89 per cent user acceptance for speed and cost-effectiveness. However, the system falters with oral

agreements or culturally nuanced disputes prevalent in developing economies, highlighting the need for human intervention where context and ethical considerations cannot be automated.^[17]

Likewise, the AI-Legal Advisor module of the e-Justice Portal of the European Commission reduced backlog of cases by 30 per cent in the pilot areas. While this tool automates a great portion of the processes for EU consumer law cases, it lacks the ability to manage multi-jurisdictional disputes, like those involving the United States of America [“USA”] Cloud Act and the General Data Protection Regulation [“GDPR”] data privacy laws, demonstrating the lack of sophistication in present-day AI.

Addressing the technical challenges in AI is very difficult, with issues of explainability and transparency extension appearing to be the most critical. Decisions made with the aid of AI that cannot be reasonably interpreted, otherwise known as the ‘black box’, do not comply with the New York

Convention requirements for due process. Methods of Explainable Artificial Intelligence [“XAI”], give the ability to create layman’s reasoning for AI decisions, termed LIME for example. In 2023, a pilot project conducted by the ICC tackling an AI arbitrator’s rationale for awarding in breach of contract case adjusted LIME to trace the logic of reasoning accounting for human legal thought. These concerns are also beginning to be addressed through regulatory frameworks.

The proposed EU Artificial Intelligence Act of 2024 requires high-risk AI systems such as arbitral algorithms to keep ‘transparency logs’ noting all decision-making steps in a format that is easily interpretable. These logs, which are similar to traditional arbitral notes, could improve accountability without sacrificing efficiency.

AI trained on legacy data risks reproducing systemic biases like procorporate or gender bias. A 2021 commercial arbitration AI tool favoured

employers in 68 per cent of employment disputes due to biased training data. Fairness audits and representative training datasets serve as protective measures. The OECD Fairness Indicators framework, adopted by the ICC in 2023 guidelines, provides methods for removing algorithmic bias. Annual independent audits ensure fair operation, enhancing AI equality.^[18]

Where human arbitrators have the ability to self-correct or learn in the moment, AI systems do not possess this ability. With the HOF, however, human verification levels are built-in to ensure the accuracy of AI generated information. The Singapore International Arbitration Centre [“SIAC”] tested these models in a pilot study in 2023 and proved it was effective. In this pilot study, a group of human arbitrators evaluated the AI preliminary awards, which decreased the number of mistakes AI made by 42 per cent. This model does not only compensate for the lack of modern technologies; it also meets the needs of people who prefer old school arbitration methods.^[19]

Integrating AI into arbitration workflows requires phased implementation to prevent disruption. Initial stages should focus on system setup and document organization, where AI excels. JAMS’ Smart Resolution Platform illustrates how systems can comply with GDPR regulations with blockchain technologies, which allow timestamped, encrypted evidence to be stored offline. Highly developed infrastructures allow for the final stages to be automated, which uses AI to produce draft decisions for reviewers; during this process, data at rest, data in use, or data in transit remain encrypted in order to maintain the privacy of the commercially sensitive data.

Federated learning and quantum computing promise to revolutionize the industry. Quantum computers run at unparalleled speeds, processing and crunching massive volumes of data in real-time. This means, in theory, AI systems can resolve complicated multi-actor disputes, containing thousands of documents, within minutes. For instance, International Business Machines reports its 2023 quantum computing trials, which achieved a 90 percent reduction in processing time for legal discovery phase data. Federated learning mitigates privacy risk by constraining an AI model within a data silo of a specific user; the AI model then gets updated when the hosting data silo is modified. To demonstrate privacy preservation, a 2023 partnership between the Hong Kong International Arbitration Centre [HKIAC] and Google involved AI learning from case law, albeit without revealing the sensitive case law to the AI.

However, the human element is the most critical. HOF appreciates the need to balance by allocating decisions to human arbitrators in the most significant and complex matters, while reserving for an AI system the role of a procedural assistant.

       IV.        Procedural Justice and Ethical Safeguards in AI Arbitration 

The integration of AI into arbitration raises important concerns regarding what is known as procedural justice, which guarantees that the methods of managing and resolving conflicts are fair, transparent, and equitable. However, the advent of automation, along with the entirely new risks it brings, is undermining procedural justice, which is vital in arbitration.

1. Transparency and the Right to Understand

Procedural justice requires that parties understand the reasoning behind decisions affecting their rights. Conventional arbitration has relied on reasoned awards, where an arbitrator gives the award by setting out the legal facts that support the conclusions reached. Many modern systems powered by AI, function as ‘black boxes’, generating outputs through opaque processes with no hope of human understanding. Opaque or unexplainable algorithms ‘black boxes’, risk violating due process under Article V(2)(b) of the New York Convention, which allows refusal of enforcement where awards offend public policy.” Achieving this goal is only attainable through the use of XAI techniques. For example, LIME was developed to explain AI decisions in way that makes sense to humans. Furthermore, regulatory approaches are being developed to alleviate these problems. Frameworks such as the EU Artificial Intelligence Act (2024) require transparency logs for high-risk AI, promoting explainability comparable to reasoned arbitral awards. A 2023 study conducted by Stanford Computational Policy Lab found that explanations produced using LIME in the context of commercial arbitration were routinely given in a very simplistic manner, and therefore were misleading. In any case, transparency mechanisms must be supplemented by manual actions to verify that they are sufficient.

• Mitigating Algorithmic Bias

The use of historical arbitral data to train AI systems creates risks of perpetuating systemic biases, such as a preference for corporate subjects over individual claimants or gender biases in employment disputes. These forms of AI bias run counter to the foundation of UNCITRAL Model Law which seeks to achieve impartiality and trust in AI arbitration. To alleviate these damages, it is recommended to follow these specific actions in response to the training data:

1. Corrective actions regarding training data – Data sets should be developed and organized to show equity across varying populations, regions, and categories of conflict. The ICC’s 2023 Guidelines on AI in Arbitration recommend the use of augmenting training data with synthetic cases to represent small businesses or non-Western legal traditions.
2. Fairness Audits – These involve the use of other frameworks, like the OECD’s Fairness Indicators, to evaluate if there is bias in the algorithm’s output. SIAC has become the first institution to conduct annual bias audits for its AI tools by outsourcing ethical decision-making. 
3. Dynamic Bias Correction: They have the ability to detect biases immediately, resulting in skewed outcomes. The proposed AI Act of 2024 for the EU called for ‘risk-based monitoring’ to be executed for greater AI systems. This means that arbitration platforms are now required to incorporate algorithms that notify human arbitrators when they are faced with undue disparity. For example, in 2023, the Hong Kong International Arbitration Centre (HKIAC) conducted a pilot project that applied real-time techniques to combat evident gender bias in employment disputes.

C. Accountability and Liability Gaps

A core challenge that persists within AI arbitration is attribution of responsibility for erroneous results. Conventional arbitration places human arbitrators at the centre of legal responsibility on account of wrongdoing based on independence and impartiality concepts. 

Efforts are being made at the legislative level to fill this void:

1. California’s AI Accountability Act (2023)^[20] requires arbitration institutions using AI to obtain insurance covering algorithmic errors.
2. The innovation provided by the HKIAC’s 2023 Pilot Program is the requirement for human arbitrators to review and sign all AIproduced awards, thereby guaranteeing observance of procedural fairness and equity.

D. Ethical Dilemmas: Efficiency v. Human-Centric Justice

The reliance on AI tools to increase productivity undermines the purpose of arbitration by turning it into a mere business deal rather than an equitable exercise. ^[21] Scholars such as Floridi have previously noted the risks of dependency on algorithms, by claiming that ‘moral imagination’, a concept most useful in handling disputes embedded in culture or new legal norms, is lost in the process.^[22]

Achieving ethical AI application alongside productivity requires delineating limits on the actions that AI can perform. The HOF distribution of functions suggests tiered decision making:

Tier 1: AI is fully in charge of all procedures, from scheduling meetings to managing relevant documents.

Tier 2: Human-AI partnership for evidence evaluation and for searching relevant cases.

Tier 3: Human arbitrators make decisions on sensitive policy issues, and in cases with morally intricate dilemmas, such as those involving human rights.

E. Global Harmonisation of Ethical Standards

Due to a lack of universal standards that govern AI in arbitration, the consequences lead to fragmentation. UNCITRAL Working Group III is drafting Model Guidelines on AI in Arbitration incorporating principles

from the OECD, IEEE, and the EU, grounded in proportionality between system complexity and dispute value. ^[23] 

• The Hybrid Oversight Framework (HOF): Balancing Innovation and Integrity

The Hybrid Oversight Framework (HOF) integrates AI efficiency with human oversight to preserve arbitral integrity. As outlined in previous sections, the legal, technical, and ethical matters are HOF’s top priority.

1. Conceptual Foundations of the HOF

The HOF is a continuation of the ICCA-Queen Mary Task Force’s 2020^[24] suggestions on the implementation of AI in arbitration, which cites the ‘human-in-the-loop’ concept, to preserve fairness while leveraging technology. The model rests on three principles: procedural equity (transparency and reason-giving), risk-based task allocation (AI for low-risk tasks), and enforceability compliance (alignment with the New York

Convention).”

• Components of HOF
  • AI-Driven Argumentation

“Under HOF, AI performs preliminary tasks such as document review and evidence labelling. For example, Brazil’s JIBOIA platform employs NLP tools to analyse claims and extract contractual clauses, reducing delays in small commercial disputes by 65%.”

1. Human Oversight Mechanism 

In 2023, SIAC implemented this model as a pilot program where all AI generated preliminary awards were subject to prompt and mandatory human intervention. 42 percent of errors were removed while sustaining efficiency, showcasing the value of hybrid workings

1. Enforcement Safeguards

To guarantee enforceability, the HOF integrates transparency logs describing algorithmic procedures, and certification standards modelled after Singapore AI Verify Framework and ISO/IEC 24028:2020 to provide assurance on bias mitigation and accuracy.

C. Implementation Roadmap

1. Phased Adoption

The HOF envisions phased adoption: initially automating administrative functions such as document sorting (already reducing LCIA costs by 30%), progressing to AI-assisted evidence analysis and procedural orders, and culminating in human-ratified award drafting, as tested by HKIAC in tradesecret disputes.

1. Technical Infrastructure

Technological integrity can be reinforced through blockchain timestamping of arbitral records, as used in platforms like Kleros, and privacy protection through encryption and federated learning in compliance with the EU’s 2021 AI-in-Arbitration study.

D. Addressing Implementation Challenges

1. Jurisdictional Fragmentation

Divergent national positions on AI personhood threaten the HOF’s global applicability. To counter this, the UNCITRAL Working Group III is considering Model Guidelines to standardise transparency and humancontrol requirements. ^[25] which would set standardised requirements of transparency and human control.

1. Training and Capacity Building

The 2023 ICC Report on Arbitrators’ Skills^[26] contains a discussion on the inadequacy of arbitrators’ ability to apply AI.  

Skill-building initiatives, such as Chartered Institute of Arbitrators [“CIArb”] AI arbitration certification courses and OECD’s policy modules for legal practitioners, are crucial for equipping arbitrators to engage effectively with hybrid systems.”

E. HOF in Action

Following JIBOIA’s 2022 integration of a human-review layer, user satisfaction rose to 92%, illustrating that hybrid review improves both pace and equity.

• Policy Recommendations and Legislative Pathways 

Integration of AI in international arbitration hinges upon the development of legal and ethical parameters that allow for the exercise of legal discretion and flexibility. The following considerations outline reforms to allow for the exercise of equitable access and for the exercise of discretion and flexibility.

A. Legislative Reforms

Within the context of international arbitration, the New York Convention is the “centerpiece” of cross-border arbitral enforcement; however, its scope is interpreted expansively, and its “requirements” give rise to “special” supplementary provisions such as annexes. Amendments essential for that scope may include:

i.The New York Convention Needs to be Modernized

Article II(2) should include provisions that recognize ‘cryptographic digital signatures as satisfactory, and blockchain timestamping as documents that refer to the ‘in writing’ criterion as stipulated within Singapore’s Electronic Transactions Act 2021’. An annex should, in the same way, argue the importance of transparency relating to AI-driven awards in constructs and log systems as evidenced in documentation required by the EU AI Act 2024.

ii.National Legislation for AI Arbitration 

National laws should require AI-arbitration systems to comply with trustworthiness and risk-management standards (ISO/IEC 24028 and ISO 31000). Liability models akin to California’s AI Accountability Act 2023 mandating insurance coverage for algorithmic errors would provide recourse and incentivise testing.

B. Intuitional Initiatives

i.Model Clauses for AI Arbitration Agreements 

Institutions such as the ICC and LCIA should include model clauses specifying the extent of AI participation e.g., procedural management or preliminary drafting and consent provisions under Article II(1) of the New York Convention. The LCIA’s 2023 Draft Clause offers a workable example.”

ii.Global Certification and Oversight 

This is to propose the establishment of an UNCITRAL AI Arbitration Oversight Board with the following powers and functions:

An UNCITRAL AI Arbitration Oversight Board could approve AI technologies, resolve cross-border enforcement conflicts, and publish annual fairness reports based on OECD and IEEE ethics standards similar to ones done by Singapore International Arbitration Centre [“SIAC”] in 2023

                                C.         Ethical and Technical Standards

i.Bias Mitigation and Fairness Audits 

External audits using OECD Fairness Indicators, as trialed in SIAC’s 2023 pilot, reduced pro-corporate bias by 22%.^[27]

ii.Human-AI Collaboration Protocols 

Capacity-building through the CIArb AI Arbitration Course and development of open-source toolkits modelled on JAMS Smart Resolution can help smaller institutions adopt affordable hybrid workflows. D. Economic and Accessibility Measures

i.Subsidised AI Infrastructure 

To ensure equitable access, bodies such as ICSID could subsidise AI infrastructure for developing jurisdictions. Pilot programmes, like Kenya’s 2023 AI-based document-management system that reduced backlog by

40%, illustrate potential gains. Scaled-down versions of Brazil’s JIBOIA

model could extend affordable small-claims resolution in emerging markets.^[28][29]

ii.Global Funding Mechanism 

An UNCITRAL AI Arbitration Fund supported by member states and private stakeholders could finance deployment of AI infrastructure in developing nations, addressing funding gaps identified by OECD studies^[30]

• Comparative Analysis of Legislative Models

Jurisdiction	Key Legislation	Focus	Impact
EU	Artificial Intelligence Act 2024	Transparency, High Risk AI	Mandates           XAI Logs for Arbitral Algorithms
Singapore	AI              Verify Framework 2022	Certification, Bias Audits	Reduced procedural delays by 40 per cent in SIAC Cases
USA  (California)	AI Accountability Act 2023	Liability, Insurance	Sets Precedents for Algorithmic error compensation

Brazil

Civil Update  2021

Code

Informal Sector Access

Enabled 50,000+ claims via JIBOIA

• Case Studies in Legislative Success

Recent initiatives show promising outcomes: the EU’s transparency mandates under the AI Act improved user trust by 30%; Singapore’s AI Verify Framework cut bias in SIAC cases by 25%; and California’s liability regime reduced algorithmic bias in employment disputes from 65% to 12% within a year

                  VII.      Case Studies: Global Implementation Efforts 

A. European Union: AI-Legal Advisor in Cross-Border Consumer Disputes

The AI-Legal Advisor is an AI module designed to handle low-value crossborder consumer disputes. It was integrated into the EU’s e-Justice Portal to tackle the backlog of cases national courts face. The AI-Legal Advisor deals mainly with matters concerning the returns of defective products and service delivery claims under EU Directive 2013/11/EU. Only claims below €5,000 are accepted into the automated system set up by the portal. 

i.Implementation 

1. Human Oversight: Personal data of those residing outside the EU poses jurisdictional complexities that compel humans to intervene in intelligent arbitration scenarios. If AI-awards are challenged in any way, automatic issuance of such awards is prohibited according to the Brussels I Regulation implemented in 2022.

ii.Challenges 

Concerns over unclear claim rejections led the European Commission in 2023 to introduce transparency logs explaining how evidence and legal precedents were applied. The same year, multilingual NLP updates enabled the system to process claims in all 24 official EU languages, resolving earlier linguistic barriers faced in Eastern Europe.

iii.Outcomes

1. Enhanced Productivity: In Germany, the average resolution time for consumer disputes improved from 12 months to 6 weeks. In 2023, there were over 35,000 cases in Germany that were resolved using AI.^[31]
2. Customer Satisfaction: According to an EU survey conducted in 2023, 92 per cent of respondents reported satisfaction with the speed and the low cost of AI mediated resolutions while 18 per cent of users reported dissatisfaction due to the lack of human involvement.^[32]

iv.Lessons Learned 

The case studies highlight that more focus should be put on the aid of AI while maintaining human supervision and providing transparency in the award giving process. 

B. Singapore: AI Verify Framework in Commercial Arbitration

i.Background 

Singapore’s ‘AI Verify Framework’, established in 2022 by the Infocomm Media Development Authority, assesses an AI system’s bias mitigation capabilities, accuracy, and level of transparency, to determine if it can be utilised for arbitration.^[33]

ii.Implementation 

1. Certification Process: The actual quality assurance procedure involves testing the explainability and other criteria of the system through Fairness Indicators and the OECD’s ISO/IEC 24028 and is done by third parties. The internationally accepted standards along with Singapore’s Trusted AI regulations under the arbitration act 2022 amendments have made it possible to grant such borders a trusted AI seal.
2. Workflow Integration: In 2023, SIAC used trusted AI systems that have been granted certified statuses for managing case documents as well as timelines in 150 cases. However, SIAC mandated that award writing had to be done by trained arbitrators.

iii.Challenges

Initial resistance from practitioners stemmed from fears of job displacement, which SIAC addressed by launching CIArb-led training programs, certifying over 200 arbitrators to oversee AI use by 2023. To safeguard sensitive maritime contracts, SIAC also adopted specialised blockchain storage with GDPR-level encryption for enhanced data security.^[34]

iv.Outcomes 

1. Effectiveness Measures: More than half of the participants in trial cases experienced a reduction in procedural delays of over 40 per cent. Document review periods were cut down from three weeks to five days.
2. Compliance: There were no disputes resulting from the AIassisted awards that were adjudicated in New York. This confirms that the framework meets international best practices.

v.Lessons Learned 

Singapore’s model demonstrates the importance of stakeholder

engagement and effective monitoring and evaluation systems.                                            

C. United States: JAMS Smart Resolution in Commercial Dispute Mediation

i.Background

In 2021, JAMS, the largest private arbitration company in the USA, released Smart Resolution, an AI-enhanced platform for negotiation and arbitration that incorporates blockchain technology and is designed for commercial disputes under $500,000. It focused on technology and information law industries.

ii.Implementation 

1. AI Negotiation Module: The system recommends settlement proposals to the parties based on award history and JAMS’s federal court’s prior decisions. The parties can either concede to the AI or ask for a human arbitrator to decide the case.
2. Blockchain Transparency: Each participant’s use of the proprietary blockchain to submit evidence and AI proposals or to accept settlement agreements is recorded with an encrypted digital signature and is time-stamped, making the evidence auditable. iii.Challenges 

1. Early concerns about the enforceability of AI settlements were resolved when the California Supreme Court in TechStart Inc. v. Innovate LLC (2022) upheld blockchain-based awards under the Federal Arbitration Act. A 2022 audit also found a 65 per cent bias favouring Silicon Valley firms, which JAMS reduced to 12 per cent by retraining its models with more representative data.

iv.Outcomes 

1. Decrease in Cost: The average costs of arbitration decreased by 50 per cent from $25,000 to $12,500, making it affordable to startups that were previously priced out from using conventional arbitration.
2. Adoption Increase: In 2023, it was reported that 70 per cent of all JAMS administered technology disputes began with Smart

Resolution. There was an 85 per cent settlement rate at the AI stages of the negotiations.

v.Lessons Learned 

The ability of JAMS to achieve its goals rests on the trust associated with blockchain as well as the use of advanced machine learning technologies. D. Brazil: JIBOIA Platform for Small Claims in the Informal Sector

i.Background

The JIBOIA Platform in Brazil, which commenced operations in 2020, aims to meet its informal economy with small claim resolution services, helping solve disputes under the value of $10,000. It was developed in partnership with CBAr – the Brazilian Arbitration Committee – integrating AI based analysis and community dispute resolution tools.

ii.Implementation 

1. Culturally Shaped: The neural linguistic processing (NLP) used in JIBOIA is trained on Brazilian informal sector oral agreements, such as those made between street vendors.
2. Community Relayers: AI produced outputs are customised by local moderators, taking into account cultural particulars like the tradition of family paying debts.

iii.Challenges 

Initially, only 60 per cent of JIBOIA’s target users had internet access, prompting CBAr and Banco do Brasil to establish kiosks in underserved regions, raising access to 85 per cent by 2023. Early bias favouring registered companies was corrected by adding 10,000 informal sector cases to the training data, improving fairness by 27 per cent.

iv.Outcomes 

In 2023, JIBOIA resolved nearly 50,000 disputes, clearing about 30 per cent of the small claims court’s pending backlog. A survey conducted by CBAr the same year reported that 94 per cent of users were satisfied with the platform, commending it for its cultural sensitivity and the speed with which it delivered resolutions.

v. Lessons Learned 

Case Study	Jurisdicti on	Key Technolo gy	Efficien cy Gains	User Satisfacti on		Legislativ e Support
EU AI- Legal Advisor	EU	NLP, Machine Learning	75        per cent Faster	92          per cent		Brussels I Regulations (2022)
Singapo re          AI Verify	Singapore	ISO/IEC 24028, Blockchai n	40        per cent Faster	88 cent	per	Arbitration Act (2023 Amendme nt)
JAMS Smart Resoluti on	USA	Predictive Analytics, Blockchai n	50        per cent Cost Reducti on	85 cent	per	California AI Accountabi lity Act
Brazil JIBOIA	Brazil	NLP, Communi ty Mediation	30        per cent Backlog Reducti on	94 cent	per	Brazilian Civil Code (2021 Update)

JIBOIA has brought attention to cultural contextualisation and grassroots partnership as it relates to AI arbitration. Other platforms in Colombia and South Africa have also adopted this model.

                                            VIII.      Conclusion 

In the development of dispute resolution systems, integrating AI within international arbitration is gaining importance and cannot be overlooked. However, this is going to need a delicate balance between advancement and values.

A. Synthesis of Contribution

1. The concept of legal validity: AI-created awards are compatible with the New York Convention and the interpretation of ‘in writing’ can be liberal as it is with digital signatures and transparency logs which

are now a part of legal practice, as can be seen in the iterations of Singapore and the EU.

• Technical feasibility: AI has the ability to improve workflow efficiency seen in platforms like JIBOIA and JAMS, however, this improvement comes with the downside of workflow opacity and bias. Much of the work to overcome these technological challenges should be directed towards LIME and federated learning, particularly the algorithmic bias and opacity.
• Procedural Justice: The Hybrid Oversight Framework (HOF) achieves a balance between efficiency and due process by integrating algorithmic accuracy with manual oversight to address Article V compliance of the Convention.
  • Implications for Stakeholders

It is essential for decision-makers to integrate the various accountability and certification frameworks, including the California AI Accountability Act alongside the UNCITRAL draft guidelines. Specific training needs to be developed for CIArb-certified arbitrators to manage hybrid workflows. System designers must include transparency and sensitivity to the system users’ cultures as the primary design factors.

• Future Directions
• Grassroots Implementation: The African Continental Free Trade Area (Af CFTA), as a regional entity, ought to be the very first implementer of AI arbitration for SMEs based on the JIBOIA model for the resolution of economic disputes.
• Ethical AI Governance: The AI systems must undergo certification and fairness audits, and problems of cross-border enforcement must be addressed by the Specialized Committee on AI and Arbitration, particularly the UNCITRAL AI Arbitration Oversight Board.

D. Final Call for Action

Evolution in arbitration will not focus on humans vs. machines as it will rather focus on the development of systems that work in tandem. The future of arbitration does not revolve around the elimination of human decision-making but rather on the optimization of human decision-making with the aid of machines. Fair and transparent systems can result in AI systems shifting in the direction of the Conventions vision of justice that is quick, impartial, and universally enforceable.

To quote the authors of the New York Convention again, arbitration aims to provide justice that is, ‘non-discriminative and executable without impediment’. In the case of AI, this is achievable with a HOF. With this vision in mind, we have taken a step closer to a world in which technology complements rather than displaces the efforts of humanity in the pursuit of justice.

HARD HATS AND SMART TECH – MODERN TECHNOLOGY IN CONSTRUCTION ARBITRATION

Alice To^

Abstract

This paper explores the impact of technological integration within construction arbitration and legal practice. The emergence of digital tools such as building information modelling [“BIM”], artificial intelligence, and cloud platforms is revolutionising processes that were previously manual in nature. Efficiency, productivity and convenience have been remarkably enhanced. The adoption of virtual hearings and digital evidence demands the development of standardised protocols to ensure clarity, fairness, and increased confidence in the wider use of modern technology. Recognising that legal professionals often lack IT expertise, the paper proposes recommendations for multi-disciplinary partnership between construction-legal and technical experts to maximise the benefits of technology in the arbitration of construction disputes. Despite the clear advantages, both the legal and construction sectors encounter resistance mostly stemming from financial constraints, knowledge gaps and concerns about technology’s impact on professional roles and values. Addressing those challenges necessitates comprehensive strategies including robust training, supportive policies, and updates to institutional procedures. Ultimately, technological advancement is positioned not as a replacement for human intelligence, but as a partner enhancing decision-making, transparency, and fairness in dispute resolution. Proactive adoption coupled with risk management and support for practitioners will

^  Ms. Alice To is a Hong-Kong based construction lawyer and arbitrator.

enable stakeholders to uphold credibility, foster innovation, and sustain the reliability of arbitration processes well into the future.

I. Complications in Construction Disputes

Construction disputes are amongst the most complex commercial conflicts due to the highly technical and detail-oriented nature of construction activities frequently convoluted by complex human dynamics and inherent uncertainties.

A typical construction project involves multiple parties at various tiers of the supply chain, operating within volumes of interdependent contracts, dealing with elaborate technical specifications, substantial documentation, and significant financial commitments. Achieving successful completion requires meticulous attention from all stakeholders. Yet, unpredictable variables and frequent changes persist. As a result, errors, delays, inaccuracies, and unforeseen impossibilities often lead to differences and disagreements that escalate to claims and disputes which are essentially inevitable.  

Most challenges in construction projects relate to the elements of time and cost. Disputes are usually in respect of delays, disruptions to progress of work, extensions of time, variations and additional work which lead to further arguments in respect of extra time and costs, liquidated damages, prolongation costs and many more other issues.^[35] Other complications may arise from design flaws, buildability concerns, unexpected site conditions, quality concerns related to workmanship, and materials, payment disputes, contract interpretation, and force majeure events amongst others. These obstacles make confrontations and debates commonplace. It is difficult to name one project that has not experienced confrontations and disputes. In

fact, it is not uncommon that even before works have even started the project is already flared up with disputes and claims.  

II. Is Arbitration the Answer to Construction Disputes?

1. General Observation

Arbitration is widely used for resolving construction disputes. Stakeholders in the construction and infrastructure sectors have long been using arbitration as a method for addressing and resolving project-related disagreements. Arbitration possesses useful features including confidentiality, flexibility, party autonomy, and certainty in the enforcement of awards which contribute to its popularity in the construction industry. Parties are usually contended that the process achieves fairness and justice, and that the time and costs involved are justified. Compared to litigation, arbitration is viewed as a more modern and accessible process to justice. It has become a commonly selected ADR method within the industry. For many years, standard construction contract forms internationally have included arbitration as the designated mechanism for final determination of disputes.^[36]

• Confidentiality

Confidentiality is a feature of arbitration that is valued across industries including the construction industry. In construction disputes, for both public and private entities immunity from publicity is important. For the public sector, confidentiality prevents the disclosure of sensitive information, reduces potential criticism and maintains public confidence in government institutions which contributes to political and social stability.

In the private sector, confidentiality protects reputations and preserves future business opportunities. For instance, in Hong Kong, maintaining

confidentiality of any news of disputes is important.³ Since the Covid-19 pandemic, financiers and project owners have become more cautious and risk adverse.⁴ Financiers in particular, monitor courts records closely to ensure that participants in projects financed by them are not involved in any legal proceedings that could affect their financial standing. When it becomes known that a participant in a project with financial assistance is subject to a monetary claim, their prospect for securing future financial support diminishes. The market has observed instances where financiers suspend or withdraw funding upon learning that a debtor is involved in legal action.

• Flexibility

Procedural flexibility is a hallmark of arbitration, particularly valuable in addressing construction disputes. Unlike litigation, arbitration does not require parties to adhere to procedural rules established by courts which may have been set years ago and are difficult to change without compelling justification. These rules are not always suitable for commercial conflicts and especially in construction disputes that often involve unforeseen complexities. Construction arbitrations are typically intricate and protracted, encompassing a wide range of highly technical issues and arguments. The volume of documentary evidence is overwhelming, and expert testimony is routinely required to assist a tribunal to master in-depth understanding of technical issues relevant to the dispute. The process is rarely straightforward. Arbitration’s procedural flexibility allows parties and tribunals the freedom to develop tailored processes that are most efficient and appropriate for the unique circumstances of each case.^[37] Adjustments to

• Arbitration Ordinance, Cap. 609, § 18(1) (H.K.).
• COVID-19 and its Impact on Project Finance Transactions, NORTON ROSE FULBRIGHT (Mar.

     2020)                                                available                                                 at

https://www.nortonrosefulbright.com/en/knowledge/publications/2e8fe68a/covid-19and-its-impact-on-project-finance-transactions.

procedural orders made by a tribunal can also be achieved when the need arises.  

• Parties’ Autonomy

Arbitration provides flexibility and accommodates the preferences of parties providing them with a wide scope of autonomy. Parties take ownership of the proceedings and are in a position to determine how they prefer their disputes to be resolved. They are able to actively participate in shaping the process and may even select and nominate the tribunal which is particularly crucial in construction arbitration. The tribunal’s legal acumen, industry-specific knowledge, extensive experience, and expertise are essential for achieving efficient, timely, and cost-effective dispute resolution.

The construction sector presents unique factual, legal, and technical complexities. Voluminous documentation, such as drawings, records, meeting minutes, correspondence, calculations of payments etc. The industry operates with    practices, jargons and communication styles that differ substantially from other industries. An arbitrator who is familiar and identify with the uniqueness in the operation and behaviour of the industry and construction personnel is in a better position to assess evidence provided by parties and witnesses. This minimises both the learning curve and related expenses.

Effective evaluation of expert testimony in construction disputes is a skillset that requires knowledge, expertise and experience in technical and industry matters. Familiarity with issues commonly arising from construction projects enables the tribunal to critically assess expert evidence and avoid misinterpretations or at times, from being misled to achieve fair and accurate decisions. The ability of the parties to select a qualified decision maker is a distinct advantage and a fundamental aspect of construction arbitration.

• Certainty in Enforcement of Arbitral Awards

Courts generally support alternative dispute resolution methods and do not usually overrule the findings of tribunals, barring exceptional circumstances. Decisions made by tribunals are commonly respected and maintained, with awards being usually upheld.

The UNCITRAL Model Law [“Model Law”] and the New York Convention on the Recognition and Enforcement of Foreign Arbitral Awards [“New York Convention”] provide for the finality and enforceability of arbitral awards. Both protocols which have gained wide international recognition provide consistency in both domestic and crossborder enforcement. The Model Law has been codified as domestic laws in jurisdictions including Hong Kong, the UK, Australia, Canada and Singapore. It helps standardise and offer a uniform approach towards the procedure of arbitration and enforcement of awards. The New York Convention strengthens parties’ abilities to enforce cross border awards in the 169 countries that have ratified it.  

Grounds for challenging arbitral awards and resisting enforcement are limited as provided under the Model Law and the New York Convention. Reasons that one could rely upon to resist enforcement include: (a) incapacity of a party or invalidity of the arbitration agreement; (b) insufficient notice of appointment of an arbitrator or proceedings, or a party’s inability to present its case; (c) issues decided ultra vires the scope of submission to arbitration; (d) improper composition of the tribunal or process exceeding agreement terms; (e) disputes not being suitable for arbitration, and (f) contravention with public policy.^[38]

Mechanisms exist for parties to challenge arbitral awards. However, the criteria for setting aside an award or resisting enforcement are stringent and

awards are ordinarily enforced except in rare cases.  For instance, in Song Lihua v. Lee Chee Hon [“Song Lihua”],^[39] a request was made in Hong Kong to set aside an award issued in China. The Court examined complaints regarding the conduct of one of the arbitrators in a three-member tribunal during a hybrid hearing. Video evidence showed that the arbitrator intermittently left the virtual hearing and his attendance at the virtual hearing appeared to be sporadic. At one point he was seen sitting in a car. He also appeared to have experienced connectivity issues, and his explanation was he was proceeding to and travelling on a high-speed rail. All those awkward incidents occurred during the oral hearing when he was a member of the tribunal performing a quasi-judicial function.  

The Court was dissatisfied with the conduct of the arbitrator being complained against and found his behaviour to be disruptive of the arbitration hearing. The Court concluded that there was no apparent justice and fairness when an arbitrator was not focusing on the oral hearing. It would have amounted to violation to the most basic notions of justice if the award was enforced and hence, enforcement was accordingly refused. 

While there have been instances where arbitral awards are not enforced, such decisions are made on a case-by-case basis. Those instances do not represent common practice in Hong Kong or similar jurisdictions. In arriving in the decision that enforcement should not be granted, the Hong Kong court emphasised that Hong Kong courts maintain a pro-arbitration and pro-enforcement approach.  

                      III.      Smart Tech for Construction Arbitration

Arbitration has been proven highly effective in resolving construction disputes. To stay at the forefront in the field of alternative dispute resolution, it is essential to embrace and adapt to ongoing technological

advancements. Given that arbitration is characterised by its flexibility and discretion, it offers greater opportunities than traditional dispute resolution methods to receive and integrate technological innovation and enhance efficiency. Nonetheless, certain aspects of construction arbitration still require improvement to simplify procedural steps that are labour-intensive and time-consuming. The following section of this paper will examine areas in which technology could make a positive difference.  

1. Electronic Document Management

Legal practitioners would remember the frequent sight of lawyers’ clerks transporting heavy boxes and pilot cases burst with legal documents and evidentiary materials through busy legal hubs and court buildings. Considerable effort was devoted to producing multiple photocopies of documents and compiling bundles of documentary evidence in lever arch files. This was part of an integral part of the discovery process. The work required was arduous, labour-intensive and inefficient.

Similarly, the document-oriented construction industry has long sought transformative solutions to document management. Historically, despite the vast quantities of documentation produced for each project, documents and records were maintained solely in hard copy and collated by hand. As a result, information was frequently lost amidst extensive paperwork, and significant time was spent locating and sorting those documents. Maintaining an organised system is critical, particularly to facilitate the retrieval of relevant evidence during disputes. The introduction of document management software including web-based and cloud storage platforms, as well as AI-driven tools has streamlined the process. Effort and inefficiencies associated with organising and keeping records were largely reduced.

While electronic document management systems [“e-DMS”] make it easier to access information, there is always the risk of a cyberattack or accidental information leaks.  It is therefore important to be aware of potential legal implications in unintended leakage of information and data. For example, where personal data which are safeguarded under various legislative frameworks are included in project documents, users of e-DMS should be vigilant and take additional steps to protect such personal data in order not to fall short of legal requirements. The EU General Data Protection Regulation [“GDPR”],^[40] for instance, aims at preventing abuse and misuse of personal information has a wide geographical application. To ensure compliance of requirements such as those under GDPR, it is recommended that personal data relating to staff and personnel be stored separately from other project records and be encrypted to restrict access and maintain confidentiality.

• E-Discovery

Various measures have been implemented to update and simplify the process of document discovery. E-discovery was introduced to facilitate the efficient storage, retrieval, and sharing of substantial volumes of documents on common platforms. The Chartered Institute of Arbitrators [“CIArb”] was amongst the first institutions to introduce e-discovery by issuing the ‘Protocol for E-Discovery’ in International Arbitration in October 2008. This Protocol requires parties to discuss at an early stage the preservation and disclosure of electronically stored information and to reach agreement on the scope and method of electronic discovery. The tribunal is responsible for determining, prior to the initial preliminary meeting, whether e-discovery would be required. The tribunal is also tasked to issue directions on the types and scope of electronic evidence to be disclosed, preservation methods, and logistical matters related to e-discovery. The

Protocol also limits the scope of discovery to documents that are both relevant and material.

In 2010, the IBA Rules on the Taking of Evidence in International Arbitration were amended to broaden the definition of ‘documents’ to encompass writings and communications recorded electronically, as well as audio, visual, and other formats. Electronic documentary evidence since then has become admissible under the revised IBA Rules. In parallel with the CIArb Protocol, the IBA Rules aim to prevent excessive or unnecessary disclosure by requiring parties to identify specific files, search terms, and methodologies to conduct searches efficiently and economically.

Over time, various jurisdictions have developed jurisprudence providing guidance on the utilisation of e-discovery. For instance, in email disclosure disputes, Hong Kong courts have explained that extra care should be taken by parties to avoid repeated disclosure of identical emails since emails could be proliferated, widely distributed or attached multiple times during ediscovery.^[41] In Singapore, the case of Breezeway Overseas Ltd & Anor v. UBS AG & Ors,^[42] established that a party declining to provide an electronic document could be liable for additional costs incurred by the requesting party to obtain access to the document.  

In Digicel (St Lucia) Ltd & Ors v. Cable & Wireless plc & Ors,^[43] an English Court highlighted the importance of early discussion among parties regarding potential issues in searching for electronic documents, particularly agreements on keyword searches if they are to be employed. In West African Gas Pipeline Company Limited v. Willbros Global Holdings Inc,^[44]the English Court held that parties must ensure electronic documents are not

duplicated and that parties failing to perform de-duplication processes may be ordered to bear the other party’s costs.

C. Virtual Hearings

Since the outset of the Covid-19 pandemic, there has been a significant increase in the use of virtual hearings in arbitration. In response to this development, arbitration institutions have revised their rules to better accommodate virtual proceedings. The LCIA Arbitration Rules 2020, for example, specify in Article 14.3 that communications between parties and tribunals may occur via conference call, videoconference, or other communication technologies. Article 14.6 further authorises tribunals to enhance efficiency and expedite progress by technological means. Similar provisions can be found in Article 26(1) of the ICC 2021 Arbitration Rules which enables hearings to be conducted either in person or remotely via accepted communication methods. Under Articles 14.3, 32.5, and 39.2 of the SIAC Rules 2025, case management conferences and hearings can be held in-person, in hybrid format or through electronic communication technologies, such as, videoconferencing or teleconferencing. Although the HKIAC Arbitration Rules do not explicitly refer to virtual hearings, the 2024 Administered Arbitration Rules direct in Article 3 that tribunals should adopt procedures conducive to avoiding unnecessary delays or expenses, including consideration of technology use. This implicitly permits virtual hearings as a means of ensuring procedural efficiency. Notably, recent HKIAC statistics indicate that in 2024, 39% of hearings hosted by the institution were fully or partially virtual.^[45]

Virtual hearing rooms present a practical solution for construction arbitration given the high mobility of construction professionals and personnel who frequently relocate from project to project. By the time a

dispute is scheduled and ready for an oral hearing, relevant witnesses may have since moved elsewhere for work making it challenging to convene all participants in one physical location. Additionally, when a tribunal comprises multiple members, each selected for specific technical expertise it can be difficult to assemble qualified arbitrators from a single location given that the arbitrators may be based in different regions. Virtual hearings facilitate the participation of all individuals regardless of geographical barriers, reduce travel requirements which is in line with environmentally sustainable practices, generate cost savings and provides convenience in scheduling conferences and hearings.

The adoption of virtual hearings has generally been met with enthusiasm by users and legal practitioners. Nevertheless, certain legal concerns have arisen, particularly regarding the potential impact on the enforceability of arbitral awards as reflected in cases such as Song Lihua. One frequently raised question is whether remote hearings risk depriving parties of the right to reasonably and adequately present their case, being a valid ground for refusing enforcement.

This issue was considered before the Hong Kong Court in Sky Power Engineering Ltd v. Iraero Airlines JSC.^[46] Sky Power sought enforcement of an arbitral award rendered under the LCIA 2014 Arbitration Rules. The parties originated from different Russian cities and the arbitrator was based in London. Initially, the arbitrator ordered a hybrid hearing, with factual witnesses and counsel attending in-person at a Moscow location, and expert witnesses participating remotely, while she herself joined virtually. However, after contracting Covid-19, in light of the safety concerns expressed by Sky Power regarding travel to Moscow, the arbitrator amended the procedure to a fully virtual hearing.

The Respondent, Iraero Airlines opposed enforcement on the ground that its position was prejudiced by the remote nature of the hearing. It claimed

difficulty in assessing the Applicant’s witnesses’ demeanour and authenticity of the oral evidence given remotely, thereby it submitted it had lost its opportunity to adequately present its case. The judge ruled against all those submissions and affirmed that Article 19.2 of the LCIA Arbitration Rules expressly allows hearings to be conducted virtually. By ordering a remote hearing, the arbitrator acted fairly and impartially in determining how the hearing was to be conducted. Accordingly, the application for setting aside the arbitration award was rejected.  

D. Building Information Modelling (BIM)

BIM is a 3D digital model that integrates and displays all major components and procedures throughout the lifecycle of a construction process and allows for interaction amongst stakeholders. Effective collaboration and coordination within the project team comprising of construction professionals, consultants, contractors and owner of the project are necessary for the implementation of BIM. The model captures information from planning and design through construction, operation, and maintenance phases. A 3D model contains more data than 2D drawings.  For example, internal layouts could be more clearly represented and integrated within the 3D model than in a flat drawing. Each party involved is responsible for providing and updating relevant data in respect of their respective parts in a project. BIM enables project stakeholders to visualise each step of the construction process. Changes can be easily made. BIM enhances transparency and supports risk management during construction. Effectively, it is a ‘digital rehearsal’ of a project.^[47]

A 3D BIM may be extended to a 4D model by incorporating time-related data. Scheduling details, such as task duration and sequence, can be added to BIM to assist with timeline visualisation, managing delays, and

monitoring progress. This is useful for optimising scheduling and coordinating tasks, such as, the fabrication and delivery of materials. In dense urban areas like Hong Kong, scheduling deliveries is critical to efficient use of limited site space. BIM supports logical and optimal sequencing of work.

In dispute resolution, digital simulations assist parties and arbitrators in understanding project elements, contractual obligations, issues, and causes of delay. Visual representations facilitate clearer communication and contribute to efficient resolution processes.

However, BIM models created after project completion may not reflect conditions as accurately as those developed contemporaneously depending on the availability and accuracy of source documents and records. It is recommended that BIM be implemented early in a project and be updated regularly to reflect what was actually happening in the project.  An as-built BIM at project completion is also recommended as it provides detailed reference data for future use and is particularly useful in the event of disputes.  

While BIM offers various advantages, challenges remain. For example, in

Colmat Construction & Engineering Co Ltd v. Minmetals Condo (Hong Kong) Engineering Co Ltd [“Colmat Construction”],^[48] the sub-contractor raised concerns about discrepancies between provided 2D drawings and 3D BIM models both of which were provided by the main contractor. Colmat submitted that it had based its work on the 3D BIM which was later instructed to be changed. In deciding that the subcontractor was liable for the additional costs and expenses in changing the completed works, the Court accepted the main contractor’s evidence that the subcontractor should have known that it was the 2D drawings which had been approved by the government and hence works should follow the approved drawings

in 2D. The Court further opined that in case of discrepancy between a 3D BIM and approved drawings, it was a duty upon the party carrying out the works (the sub-contractor in the case) to seek clarification.  The case highlighted the importance of establishing the contractual role of BIM and ensuring clear communication regarding its use in projects. Incorrect usage may undermine the intended advantages to be offered by BIM.  

                                 IV.      Challenges and Limitations

The integration of technology into daily life is now commonplace. Yet, levels of acceptance continue to vary across professions. Within the legal sector, practitioners have traditionally exhibited caution against technological advancements. There remains to be resistance to embracing technology in the field of arbitration, a core element of legal practice. Many lawyers highly value and respect tradition. Any significant changes from the traditional practice may be met with strong feelings and reluctance, particularly amongst senior members of the profession. This hesitation in adopting new solution is often due to the insecurity of departing from established practices.

A 2019 study on the attitude towards engagement of AI in the medical and legal professions (both being highly educated and specialised professions) indicates that when compared to their legal counterparts, physicians are more receptive to artificial AI and view it as a supportive resource that physicians could partner and excel with in their work. Lawyers, however, took a more conservative approach and perceived AI as having only a neutral impact on their work.^[49] Increased education and targeted advocacy are necessary to encourage legal professionals to recognise technology as a means to enhance efficiency and improve client outcomes. Thoughtful and

strategic adoption of technology is increasingly essential for sustainability within traditional fields.

Turning to the construction industry, a 2025 survey of construction firms showed that, although AI and automation are recognised as beneficial modern technologies, they remain to be under-utilised.^[50] Most respondents agreed that AI has the potential to transform project results, provide realtime market insights and predictive analytics that could help make better management decisions. Despite this optimism, only 13% of participants expressed a strong intention to invest in AI solutions within the next two years, with the majority preferring to adopt a cautious approach. The survey further revealed that over half of the companies identified were already troubled by primary concerns of rising material costs and labour shortages in their businesses and were reluctant to divert their attention from those pre-existing issues towards the adoption of innovations and AI.

Potential AI and technology users are concerned about costs, including initial hardware, software, licensing, and training expenses, as well as ongoing maintenance and upgrades. Those financial burdens can be particularly challenging for small businesses and discouraging them from adopting new technologies. Transparent pricing can help address the concerns and governments could offer support through low-interest loans or tax deductions.

The lack of knowledge and experience in using new technologies is another significant factor that contributes to the reluctance in engaging technology. Many individuals and companies may feel insecure about using technology because of the lack of knowledge and experience which can give rise to uncertainty and anxiety. They may feel pressurised facing the daunting learning curve. Without adequate and proper exposure, training and

support, users may perceive that the risks and complexity attached to the engagement of technology outweighs the potential benefits.  

Cybersecurity is also a challenge. Intellectual property, trade secrets, financial information, personal data, contracts, and other sensitive information stored online can be vulnerable to hacking risks. Maintaining cybersecurity requires robust measures which, again, may involve significant costs to ensure data integrity.

V. Outlook and Forecast

Technology has become part of life. Adopting proactive strategies can address challenges and boost trust to pave the way for greater use of technology in the construction-legal sector. Below are some suggestions to encourage more use of technology:

1. Training and Capability Building

As indicated by the recent surveys referred to above, there is an ongoing need for enhanced training and capability development within the field. Given the integral role of technology in dispute resolution, it is essential to provide targeted and customised education for arbitrators, counsel, construction professionals and related personnel regarding technological applications in this context. Training programmes should be tailored to address the specific requirements of arbitrators, legal practitioners, and participants in arbitration and other dispute resolution mechanisms, and should be provided regularly to show commitment to improve. In addition, symposiums, conferences, and workshops are highly recommended to facilitate interactive learning opportunities and to create opportunities for participants to exchange insights and experiences in the use of technology. Comprehensive training should cover all digital tools relevant to both the legal sector and the construction industry.  

Ongoing education, training, and advocacy are essential to reinforce the significance of technology in arbitration proceedings to ensure that stakeholders remain informed and proficient in its application.

• Updating arbitration rules

Technology is developing and transforming rapidly. Arbitration rules should be regularly updated to accommodate changes in technology.  Institutions must keep their arbitration rules up-to-date in order not to fall behind the progress of technological development and to avoid their rules becoming obsolete. Continual effort is required to regularly modernise arbitration rules to reinforce trust by demonstrating a commitment to accommodate the dynamic needs of the construction and legal sectors in the use of modern technology.  

Hot from the press is the Chartered Institute of Arbitration’s Guidelines on the Use of AI published on 5 September 2025.  The Guidelines aim at enhancing the efficiency in the use of AI by arbitrators, parties, their advisers and other participants. It highlights and explains the potential risks in the use of AI in arbitration which may lead to complications in enforcement of awards. Recommendations are provided to enhance the use of AI by arbitrators and parties. The power for allowing and regulating the engagement of AI in arbitration proceedings remains with the arbitrator whilst parties still enjoy a certain degree of autonomy.  

Similar guidelines and procedural rules customised for the use of technology and AI are encouraged to promote transparency and procedural uniformity in arbitration proceedings. Industry-focused protocols should also be introduced to accommodate the differences and uniqueness of disputes that arise out of different industries. An example could be directives for the use of BIM in construction disputes or other electronic means and presentations for patent disputes.  

• Development of Standardised Protocols

Incidents such as the conduct of the arbitrator in Song Lihua illustrates the need for clear and standardised protocols governing virtual hearings. Welldefined procedures would establish uniform guidelines for the conduct of virtual hearings. The receiving and use of digital evidence should also be included in the protocols. Implementing standard practice would enhance confidence in and promote more frequent use of virtual hearings which offer cost-savings, convenience, efficiency and sustainability.

• Partnership between construction-legal and IT experts

Legal practitioners and arbitrators seldom also possess expertise in information technology. As technology becomes increasingly integral to arbitration and its advancement accelerates rapidly, the engagement of IT experts is essential. Partnering with technical professionals is necessary to maximise the effective use and impact of technology within arbitration. Multi-disciplinary teams are expected to become a standard practice as arbitration processes continue to evolve with technological developments.

                                               VI.      Conclusion

The construction and legal sectors are undergoing significant transformation due to the integration of technological tools including BIM, artificial intelligence, electronic document management software, cloud platforms, e-discovery, and virtual hearing rooms. Processes that were previously executed manually are now managed electronically. Automation enhances efficiency and productivity. Technology has introduced the level of convenience and effectiveness that was never experienced prior to this era of modernisation.  

Nonetheless, human elements remain essential to the successful use and application of innovations. The case of Colmat Construction highlights the importance of human discretion in leveraging BIM effectively for favourable outcomes. Concerns regarding technology replacing human involvement are therefore unnecessary as human judgment continues to play a vital role in the success of digital advancements.

In the legal profession, especially in arbitration, adopting digital technologies has been a crucial progression rather than a mere convenience. However, despite evidence from surveys and studies showing the benefits of technology, hesitation and reluctance persist in the sector. Legal practitioners traditionally rely on their analytical skills grounded in evidence, expertise, and experience. It appears to be a common concern within the profession that wider engagement of technology may diminish their professional capabilities and values.   

The construction industry is more accustomed to technological integration. While many professionals in this field are optimistic about automation’s transformative potential, hesitation nevertheless remains, which has resulted in the phenomenon that only a small proportion of firms in the industry have made adequate investments in technology. Key obstacles include financial limitations, knowledge deficiencies, and concerns about steep learning curves – challenges that are particularly noticeable amongst smaller organisations.

To address the gaps and enable markets to fully achieve and benefit from technological integration, a comprehensive and multi-dimensional strategy is required. This should comprise significant investment in robust training and educational programmes for all stakeholders from arbitrators and legal professionals to construction managers and others. The development and adoption of standardised digital protocols will also help. Additionally, supportive government policies and collaborative efforts between legal and technical experts are vital. Arbitration institutions should also consider updating their rules and procedures to prioritise the use of tech and to ensure that all participants are prepared to engage with virtual hearings and digital evidence. In addition, providing support to legal practitioners for their sentiments of uncertainty and insecurity about the potential impact of technology on their professional value could further facilitate the adoption of technological solutions.

Looking into that crystal ball to project to the future, the effective adoption of technology in construction arbitration will continue to play an essential role in resolving and deciding the intricate technical and contractual issues in construction disputes.  It is anticipated that with continuing growth in technology the path to justice could be expedited at an even higher speed with a higher degree of accuracy and fairness in the process. Stakeholders who proactively embrace digital innovation with careful risk management policies will be best positioned to excel and uphold the credibility and value of their respective professions.

Ultimately, it is essential not to perceive technological advancement as a threat to replace human intelligence. Rather, it serves as an influential tool to be leveraged by and partner with individuals. Promoting openness, collaboration, and practical skills within the construction and legal sectors to use technology in a productive and secure manner is crucial for achieving greater efficiency, transparency, and excellence. Such efforts also contribute to sustaining and enhancing the long-term robustness and reliability of arbitration processes.