In November 2025 the AAA-ICDR announced that its AI-Arbitrator is now available for use in construction disputes. In light of this, Stacy Sinclair considers the use of AI in dispute resolution and asks, are we ready for this?
Introduction
Artificial Intelligence (“AI”) continues to fuel worldwide debate, dividing global opinion and polarising society on issues of regulation and governance, data protection, accountability, ethical responsibility and technological progress. Some view AI as a potential saviour for humanity and the planet, while others at the opposite end of the spectrum fear it could ultimately lead to our destruction.
Regardless of differing perspectives, it is undeniable that AI is here to stay.
Following OpenAI’s release, or rather, the explosion of ChatGPT into the market in November 2022, AI is now a household name. AI has been democratised.[1] AI is no longer confined to technology specialists in research labs, but it is embedded into our everyday life, making advanced technologies accessible to both individuals and organisations of every size, across every sector – including law and construction.
As AI systems become more capable, their reach is extending into domains traditionally reserved for human judgment – including the resolution of disputes. AI already is assisting construction professionals, lawyers, tribunals and judges in dispute resolution by analysing vast datasets, legal research, predicting outcomes, helping to frame issues and assisting in daily drafting and summarisation tasks with unprecedented speed and consistency. What was once experimental has quietly become routine.
While these tools do not yet replace human decision-makers, they certainly are increasingly shaping the information and arguments relied on by human decision-makers. In doing so, the boundary between decision-support and decision-making itself is blurring, raising inevitable questions about what comes next.
How far away are we from the AI-Judge, the AI-Arbitrator, or the AI-Adjudicator? Are we ready for this?
Democratisation of AI
The prospect of AI-Judges, AI-Arbitrators and AI-Adjudicators is already sparking debate, but this is far from unprecedented. Major technological shifts often are met with scepticism, uncertainty and heated discussion. Yet, once accessible to the general public, we have seen how they become embedded in society, ultimately reshaping how we work, communicate, decide and deliver outcomes. For example, in recent history:
The Personal Computer (PC): In the 1980s, it was the PC that was placed into the hands of the wider public. In 1982, Times Magazine[2] declared the PC as the “machine of the year” and Marvin Minsky, an eminent MIT computer scientist commented: “The desktop revolution has brought the tools that only professionals have had, into the hands of the public: God only knows what will happen now”.[3] Another leading AI expert at the time, Nils Nilsson, said the PC, like television, can “greatly increase the forces of both good and evil”.[4] Today, AI similarly is regarded as a technology that can magnify our capacity for both progress and harm.
The Internet: In the 1990s, it was the internet. In 1994, Netscape, one of the first leading web browsers made its debut, known as the “Netscape Moment”.[5] Again, before its release, the internet was only in the hands of the privileged few. Once released, and the general public had access, it changed the way we communicate, shop, date and find partners, and even vote. It changed the way we live, work and understand our environment. In business, those who did not get onboard with the internet struggled to keep pace with those who did, and many ceased trading. Again, when released, the reaction was much the same with views of mistrust and anarchy and headlines such as “the Internet, used by millions of Americans every day, has become a virtual textbook for terrorists”.[6]
Artificial Intelligence: And in the 2020s, ChatGPT brought AI to the general public, democratising AI. We are now facing that next “Netscape Moment”. Will it play out like the PC and the internet? My bet certainly is that those who have access to AI and take advantage of it will reshape the future, and those who do not will lose out.
The use of AI in dispute resolution is simply the latest chapter in this long-running story of technological evolution.
Use of AI in dispute resolution
AI is reshaping dispute resolution by enhancing the speed, accuracy and consistency across the lifecycle of disputes. From early case assessment and case preparation through to conducting hearings and trials and decision-making, the tools available have the capability to analyse vast amounts of information, surface critical insights in real-time, and support lawyers and decision-makers in producing clearer, more robust and more efficient outcomes.
Each stage of the dispute lifecycle itself has various activities that can be supported through AI, for example:
- advanced legal research engines rapidly locating relevant authorities;
- automated chronology builders extracting events from large datasets;
- drafting assistants for pleadings, submissions and correspondence;
- summarisation tools for distilling lengthy reports, witness accounts and expert materials;
- document review platforms which categorise and cluster datasets and identify relevance, privilege and anomalies;
- outcome-prediction models and strategic assessment tools to test positions, quantify risk and refine case tactics;
- tribunal-selection tools which draw on historical data to identify decision-makers whose expertise and past reasoning align with the dispute; and
- real-time technology which generates summaries and extracts key points from evidence as it emerges during the course of a trial or hearing, providing rapid cross-referencing across the document set.
Together, these technologies have the potential to create a more informed, efficient and strategically sophisticated dispute resolution process. Of course, the successful use of any AI system depends on navigating a range of challenges and risks, particularly:
- defining appropriate AI governance, strategies and policies;
- harnessing data and addressing issues of quality, availability, privacy and security;
- ensuring human-centred, responsible AI;
- driving adoption and change;
- building capacity and engaging talent; and
- enabling digital access, inclusion and equality.
Each of these must be considered carefully to enable the safe and effective integration of AI into dispute resolution, in a way that is responsible and ethical, supporting people and enhancing trust. Failure to do so results in systems that lack proper human oversight and therefore increased risk of error and exposure to serious consequences when outputs are relied on without review – including the well-known dangers of hallucination and misinterpretation.
In 2025, we continued to see a number of cases in England & Wales exposing such failures, a few of which include:[7]
- Zzaman v Revenue and Customs Commissioners (April 2025): The litigant-in-person confirmed his use of AI to prepare the statement of case, which cited cases that did not provide authority for the propositions advanced.[8]
- Bandla v Solicitors Regulation Authority (May 2025): Again, a solicitor cited several authorities that were fake or did not exist.[9]
- R (Ayinde) v London Borough of Haringey, Al-Haroun v Qatar National Bank QPSC (June 2025): In the Ayinde case, counsel relied on non-existent cases. In the Al-Haroun case, the claimant’s solicitor relied on fictitious cases, as he did not verify his client’s research which had been AI-generated.[10]
- Ms (Bangladesh) v Secretary of State for the Home Department (August 2025): A barrister included a false case citation produced by ChatGPT, without checking its authenticity.[11]
- Victoria Place Flats et al v Assethold Limited (October 2025): Following his own investigations with M365 Copilot, the judge concluded that a litigant-in-person provided an invalid, AI-hallucinated case citation, as well as submissions that were based on hallucinated legal points.[12]
AI & Judicial Decision-Making
In 2025, we also saw the rise in the use of AI to assist judicial decision-making, as well as a call for a debate on the use of AI to automate certain aspects of judicial decision-making.
AI assisting in decision-making
In February 2025, at the LawtechUK Generative AI event, Sir Geoffrey Vos, Master of the Rolls, encouraged the use of AI:[13]
“…there are three excellent reasons why all lawyers and judges should embrace AI: those we serve are using it. It will make what we do available to more people, more cheaply, and allow us to do necessary things more quickly, and it will be at the centre of the future work of lawyers, when claims are all about when AI has been used for the wrong things, and AI ought to have been used but was not used.”
In April 2025, the UK courts rolled out Microsoft ‘Copilot Chat’, a secure and private tool in the eJudiciary platform, encouraging judicial office holders to make use of generative AI.[14] The Courts and Tribunals Judiciary also updated their publication “AI Guidance for Judicial Office Holders”, highlighting potential use cases.[15]
By July 2025, there was a general sense, among at least a handful of the judiciary, of “opportunity regarding the use of AI for various tasks within the judiciary, particularly to improve efficiency and access to justice.” This included 12 UK judges, five of which are members of the UK Supreme Court.[16]
In August 2025, in the case of Evans v HMRC, Judge McNall expressly stated that he had used AI in the production of his decision, why he did so, and how. The judge primarily used AI to summarise documents:[17]
“This application is well-suited to this approach. It is a discrete case-management matter, dealt with on the papers, and without a hearing. The parties’ respective positions on the issue which I must decide are contained entirely in their written submissions and the other materials placed before me. I have not heard any evidence; nor am I called upon to make any decision as to the honesty or credibility of any party.”
In October 2025, in the case of Victoria Place Flats et al v Assethold Limited, Judge Gethin also used Microsoft Copilot to investigate the Respondent’s submissions.[18]
And at the end of October 2025, the Courts and Tribunals Judiciary published a further revision “AI Guidance for Judicial Office Holders”, again promoting the use of AI and setting out key risks. Notably, this latest version is more risk-focused and security-aware, highlighting such issues as “white text”.[19]
AI automating decision-making
It is one thing to use AI to assist in the decision-making process, and it is an entirely different proposition for AI to automatically make the decision itself. Given the rapid rate at which technology is developing, and the acceptance already of AI to assist in decision-making, provided adequate safeguards are in place with human oversight and responsibility, how far away are we from certain decisions being entirely automated by AI? Which categories of disputes, or administrative processes, if any, can safely be handed over to AI, and which must remain firmly under human control and judgment?
Over the past year, we have seen a growing debate around these questions. Ethical, legal, and practical considerations are surfacing, reflecting a broad recognition that the path toward automation must be navigated deliberately and cautiously, shaped by transparency, accountability, and public trust.
At the end of 2024, Lord Justice Birss noted that AI has potentially transformative implications for justice:[20]
“Looking further into the future, one could imagine that AI may very well be able to assimilate much larger quantities of data than a normal human judge. One could then be faced with the situation in which an AI system might be a better decision maker than a human being in those circumstances…
The question will then become an important ethical and human rights based one – in which we need to decide whether there are decisions we are prepared to have made by AI, and which decisions should remain the preserve of human beings. One could imagine for example that a decision relating to children and whether someone had committed a crime might be one where we wish to maintain human decision making. On the other hand, one might imagine that a large number of small money claims or some other similar kinds of case, might be more efficiently done by AI, in the first instance. There could then be a right of appeal to human judges after the event.”
The Master of the Rolls, Sir Geoffrey Vos, is also at the forefront of this discussion, consistently challenging the profession to confront the practical, ethical, and constitutional implications of automated justice.
On 15 October 2025, Sir Vos asked, the “big question is what should it be used for?”[21]
Sir Vos said the answer is “difficult and potentially troubling” … “Nobody can tell me why it should not be used to assess, for example, the damages to be awarded in a personal injury case.” That would cut the time taken for a ruling from years to minutes. “So acknowledging that some decisions may be taken by AI, why should we balk at its use more widely?”
Acknowledging various issues, including AI’s lack of ability to mimic human emotion, empathy and insight, Sir Vos called for “a serious debate now to consider what human rights people should have in the light of more capable AI”.
At the International Bar Association conference in Toronto in November 2025, Sarah Sackman MP also called for a debate on AI-judges.[22]
This call for deeper debate is timely, given AI decision-making is now starting to emerge in practice.
AAA-ICDR AI-Arbitrator
On 3 November 2025, the American Arbitration Association – International Centre for Dispute Resolution (“AAA-ICDR”) announced that its AI-Arbitrator is now available for use in two-party, documents-only construction disputes.[23]
The AAA-ICDR developed the AI-Arbitrator in collaboration with QuantumBlack, AI by McKinsey, and trained it on over 1,500 construction awards and refined it with expert-labelled examples.
This marks a genuinely ground-breaking moment for the dispute resolution community. For the first time, an arbitral institution has deployed an AI decision-maker to determine real-world construction disputes, moving beyond pilot projects and theoretical demonstrations. It represents a fundamental shift in how disputes can be resolved and signals the beginning of a new era in which AI may become an integral component of mainstream arbitral practice.
How does it work? Once the parties have agreed to use the AI-Arbitrator, the parties submit their claims and evidence to the AI-Arbitrator for processing. The parties then validate that the AI-Arbitrator has accurately summarised their submissions. Once the AI-Arbitrator drafts the proposed award, a human AAA-trained arbitrator reviews, finalises and issues the award. The human arbitrator is ultimately responsible for the award, not the AI-Arbitrator.[24]
What is the benefit? The AI-Arbitrator has the potential to drastically reduce the time it takes to reach a decision, and therefore lower the parties’ costs. The AAA-ICDR are reporting that early testing shows 20-25% faster resolution times and 35% or greater cost savings. In addition, by providing AI technology to the human arbitrator, increased data-driven reasoning is now possible. Given the vast amount of data in construction, equipping human arbitrators with advanced analytical tools enables them to process evidence more comprehensively and consistently in a controlled environment.
What are the risks? As with any new innovation, the AI-Arbitrator introduces risks and challenges that the industry likely will need to confront. In addition to the usual risks which must be considered with any use of an AI system (hallucinations, transparency, bias or errors in the training data, explainability, etc), questions around due process and enforceability may arise. Perhaps less so as the parties have agreed to the process in AAA-ICDR arbitrations; however, one can image challenges in any event, or in other contexts. See for example the current case of LaPaglia v Valve Corp, in a US District Court, where LaPaglia is arguing that the arbitrator “outsourced his adjudicative role to Artificial Intelligence”, therefore undermining the validity of the award.[25]
Whilst the AAA-ICDR AI-Arbitrator is not fully autonomous, and is more akin to AI-assisted arbitration, this clearly is setting the direction of travel for how technology will shape future decision-making processes. It signals that the integration of AI into core procedural functions is no longer speculative, but an emerging reality the industry must actively prepare for.
AI-Adjudicator
What are the prospects for an AI-Adjudicator? The release of the AI-Arbitrator suggests that the AI-Adjudicator might soon become a reality too.
Whether AI-assisted adjudication or an autonomous AI-Adjudicator, this must be on the horizon, particularly as many construction adjudications, on the whole, already tend to be documents-only. Furthermore, the adjudication process generally is not the “last stop shop”. Adjudication decisions are binding unless or until revised in arbitration or litigation. Perhaps this could lend itself to an AI-Adjudicator determining the dispute in the first instance, followed by a human judge or arbitrator if a party is not satisfied?
Alternatively, in an AI-assisted adjudication this would allow human adjudicators access to AI tools to interrogate evidence, process and analyse any large data sets in submissions, and generally assist with administrative or drafting tasks, in a safe and confidential platform, in the short timeframe of a 28-day adjudication.
Is the technology advanced enough for this? Is the industry ready for this?
Either way, doing nothing is not an option. We need to advance the debate around the use of AI in decision-making and address issues of trust, transparency, ethics and human rights. In my capacity as Vice-Chair of the Technology and Construction Solicitors’ Association (“TECSA”), a leading Adjudicator Nominating Body (“ANB”), I certainly can say, watch this space!
Conclusion
As AI systems become more integrated into dispute resolution, they offer significant benefits such as increased efficiency, accuracy, and cost savings. The call for debate around the AI-Judge, along with the introduction of the AAA-ICDR AI-Arbitrator, marks a pivotal shift, suggesting that AI could soon play a central role in decision-making processes.
However, the industry must address concerns related to transparency, bias, and the ethical implications of automating judicial decisions. The successful integration of AI requires careful consideration of governance, human oversight, and the establishment of robust safeguards to ensure responsible and ethical use.
Regardless of whether you think AI is here to save us or destroy us, it is imperative that we demystify and deepen our understanding of AI and learn how to leverage its capabilities to increase access to justice and create meaningful, real impact in a manner that is safe, responsible and ethical.
[1] https://www.ibm.com/think/insights/democratizing-ai
[2] https://content.time.com/time/covers/0,16641,19830103,00.html
[3] Friedrich, O (1983), “The Computer Moves In (Machine of the Year).” Time, January 3,15, as referenced in Frey, C.B. (2025), “How Progress Ends: Technology, Innovation and the Fate of Nations”, Princeton University Press.
[4] https://content.time.com/time/subscriber/article/0,33009,952176-2,00.html
[5] Thomas R, Zikopoulos P, Soule K (2024), “AI Value Creators: Beyond the Generative AI User Mindset”, O’Reilly Media, Inc.
[6] A Reuters headline noted in Schwartz J (1996), “Restricting the Internet Is Not the Way to Sem the Tide of Terror,” The Washington Post, August 5, cited in Frey C.B. (2025), p 370.
[7] For a tracker of “hallucination” cases, see: https://naturalandartificiallaw.com/.
[8] Zzaman v Revenue and Customs Commissioners [2025] UKFTT 539 (TC).
[9] Bandla v Solicitors Regulation Authority [2025] EWHC 1167 (Admin).
[10] R (Ayinde) v London Borough of Haringey, Al-Haroun v Qatar National Bank QPSC [2025] EWHC 1383 (Admin).
[11] Ms (Bangladesh) v Secretary of State for the Home Department [2025] UKUT 00305 (IAC).
[12] Victoria Place Flats et al v Assethold Limited, HAV/00HP/LSC/2024/0523, First-tier Tribunal.
[13] https://www.judiciary.uk/speech-by-the-master-of-the-rolls-at-the-lawtechuk-generative-ai-event/
[14] https://www.artificiallawyer.com/2025/04/24/uk-courts-roll-out-microsoft-copilot-for-judges-update-genai-rules/
[15] https://www.judiciary.uk/wp-content/uploads/2025/04/Refreshed-AI-Guidance-published-version.pdf
[16] Solovey E, Flanagan B, Chen D (2025), “Interacting with AI at Work: Perceptions and Opportunities from the UK Judiciary”, CHIWORK ’25: Proceedings of the 4th Annual Symposium on Human-Computer Interaction for Work, Amsterdam, Netherlands, June 2025, Association for Computing Machinery, US, https://dl.acm.org/doi/10.1145/3729176.3729192.
[17] VP Evans (as executrix of HB Evans; deceased) & Ors v The Commissioners for HMRC [2025] UKFTT 1112 (TC).
[18] Victoria Place Flats et al v Assethold Limited, HAV/00HP/LSC/2024/0523, First-tier Tribunal. See paragraphs 82-97.
[19] White text is “text formatted to be invisible to human readers (e.g. white font on a white background) but still detectable by computers. It can be used to manipulate search engines or large language models by embedding hidden instructions or keywords”.
[20] Birss LF (2024), “The Impact and Value of AI for IP and the Courts”, https://www.judiciary.uk/the-impact-and-value-of-ai-for-ip-and-the-courts-a-speech-by-lord-justice-birss/.
[21] https://www.lawgazette.co.uk/news/debate-must-start-now-on-ai-judicial-decision-making-says-master-of-the-rolls/5124769.article
[22] https://www.lawgazette.co.uk/news/iba-2025-justice-minister-calls-for-debate-on-ai-judges/5125004.article
[23] https://www.adr.org/press-releases/aaa-icdr-ai-arbitrator-now-available/
[24] https://www.adr.org/ai-arbitrator/
[25] LaPaglia v. Valve Corp. (S.D. Cal. No. 3:25-cv-00833); https://svamc.org/us-court-considered-whether-to-vacate-an-award-because-of-arbitrators-alleged-use-of-ai/.
