The Changing Face of Construction Disputes: Reforming Arbitration and Harnessing AI

Introduction

Construction arbitration has long been sold as the sensible alternative to litigation for large, technically complex projects: private, flexible, enforceable across borders, and resolved by tribunals with genuine sector expertise. That reputation is now under real strain. In 2024, Global Arbitration Review convened a high-profile roundtable to ask whether international arbitration had “lost its way”, chaired by the well-regarded arbitrator J William Rowley KC.¹ His answer was blunt: the process has become bloated, with pleadings, witness statements and expert reports that have grown far beyond what any dispute actually requires.

That criticism will feel familiar to anyone who has run a substantial construction arbitration in the last decade. According to a widely cited 2015 ICC report, party costs — legal fees and expert fees combined — account for roughly 83% of the total cost of arbitration, with the tribunal and institution absorbing only a small fraction of the bill.² Layer on top of that hourly rates at leading firms that have, by some estimates, risen by as much as 40% over the past five years, and it becomes easy to see why arbitration users increasingly describe the process as prohibitively expensive, particularly for disputes of modest or moderate value.

This article looks at two things together, because they are now inseparable: the changing shape of construction dispute resolution itself, and the arrival of artificial intelligence as a genuine tool — not a gimmick — for controlling the time and cost of that process. Construction businesses that understand both are far better placed to protect their commercial position.

Is Arbitration Still Fit for Purpose?

There remains much to commend arbitration for construction disputes: confidentiality, the ability to appoint tribunal members with real technical and legal sector expertise, and the finality and cross-border enforceability of an award. Arbitration is also, realistically, here to stay across the Gulf and other major construction markets — it is deeply embedded in regional legal frameworks, and governments have invested heavily in arbitration centres as part of wider efforts to attract international investment.

The genuine question is not whether arbitration should exist, but whether the way it is currently run serves the parties who actually use it. The consensus emerging from commentators, funders and users alike is that it does not, at least not consistently. Christopher Bogart of Burford Capital has argued that the industry needs to rediscover the discipline that originally made arbitration attractive: a genuinely faster and cheaper alternative to the courts, rather than a private substitute for exactly the same delay.³

The Shape of the Change

Three shifts are becoming visible across the industry, and construction businesses that get ahead of them are better placed to control their own dispute costs rather than simply absorbing whatever the process throws at them.

The first is a renewed interest in shorter-form and consensual processes — mediation, and in particular adjudication — even in jurisdictions where these are not yet embedded as a mandatory first step. Standard FIDIC contracts already provide for adjudication, but in the Gulf it is still often treated as an unnecessary extra layer on the way to arbitration. In practice, adjudication has proved useful even where enforceability of a temporarily binding decision remains unresolved, because it tests the parties’ respective positions early and frequently narrows or resolves the dispute long before a full arbitration would otherwise conclude.

The second shift is towards more disciplined case management. Tribunals have historically been reluctant to constrain unco-operative parties for fear of exposing an eventual award to procedural challenge. In practice, successful challenges to arbitral awards remain relatively rare. That gives tribunals more room than they often use to actively manage volume: narrowing issues early, limiting the length of submissions and expert reports, and refusing to let disclosure become an open-ended exercise.

The third is the arrival of artificial intelligence as a practical tool for controlling cost and time — not merely as a talking point at conferences, but as something genuinely capable of changing how a construction dispute is prepared and run.

AI Enters the Picture

There is no realistic prospect of reverting to a pre-AI landscape in dispute resolution. AI-driven tools are already increasing efficiency across document review, legal research and predictive analytics, and are steadily reducing the time needed for tasks that were traditionally labour-intensive. Used properly, AI has genuine potential to improve access to arbitration and litigation, rather than being simply a cost-saving exercise for the lawyers using it.

There is a fair objection to raise here: does AI simply entrench existing inequality, by giving the party with the deepest pockets access to more advanced proprietary systems, richer datasets and bespoke integrations that smaller opponents cannot match? Some law firms are investing heavily in proprietary technology. That may benefit those firms commercially, but it does not automatically translate into value for the client — developing proprietary systems is expensive, and those costs tend to be passed straight through.

A more constructive approach is to help construction clients assemble a tailored toolbox of AI solutions already available in the market, rather than paying to fund a law firm’s proprietary system. That toolbox should be built around the specific dispute: its complexity, its data volumes, the case strategy and the budget available. Off-the-shelf applications for legal research and e-discovery are now genuinely capable of supporting even complex, multi-party construction arbitrations, without the client bearing the cost of bespoke development.

Where AI Adds Real Value

Used well, AI can materially improve decision-making and predictability across the life of a dispute, in ways that go well beyond simply reading documents faster.

Early claim evaluation is the clearest example. AI-assisted review of correspondence, programme data and site records allows parties to assess the strengths and weaknesses of a case at a much earlier stage than a traditional manual review would allow, which in turn supports genuine early strategising — identifying the likely pressure points and the optimal approach before positions harden. With that improved early insight, clients are better placed to make informed decisions on funding: whether to proceed, to settle, or to pursue third-party funding, and on what terms.

AI also supports greater certainty of both outcome and cost. By identifying patterns across prior decisions and comparable disputes, it can improve risk assessment and support more accurate cost forecasting, which in turn opens the door to more creative pricing models — fixed fees and conditional arrangements become far easier to price with confidence once the scope and likely trajectory of a matter can be assessed with greater reliability.

Perhaps most significantly for construction disputes specifically, AI has real potential to support detailed, evidence-based, document-linked causal analysis — the kind of delay and disruption narrative that has traditionally required weeks of manual cross-referencing between programme, correspondence and site records — at a much earlier stage in the life of a dispute. Combined with early expert input, this gives decision-makers a genuinely evidence-based picture of exposure and entitlement while there is still time to use it strategically.

The Risks That Come With It

None of this should be adopted uncritically. Recent guidance from the arbitration community sets out the risks that come with AI in a dispute resolution context: hallucinated or inaccurate outputs, cybersecurity vulnerabilities and data breaches, fabricated or manipulated evidence including deepfakes, bias embedded within underlying systems, unequal access to advanced tools raising genuine inequality-of-arms concerns, and the risk that over-reliance on AI gradually erodes independent legal judgement.⁴

There is also a more subtle risk specific to arbitration. Increased reliance on AI may simply shift the battleground away from the underlying merits of a case and towards satellite arguments about the robustness and reliability of AI-generated output itself. The international arbitration community has no shortage of incentive to devise new ways of undermining an opponent’s case, and the reliability of AI-generated evidence and analysis is a rich new seam for exactly that kind of argument. Any cost savings achieved through AI adoption could, in principle, be offset by entirely new categories of dispute about the AI itself — its training data, its provenance, and the integrity of its outputs.

It follows that AI does not remove the need for high-quality legal and technical judgement in construction arbitration — it potentially relocates where that judgement is applied. What should change is the balance of work: less time spent on the mundane, repetitive tasks that AI performs well, and more time spent by experienced practitioners shaping strategy and interrogating AI output, rather than generating it from scratch.

What This Means in Practice

These risks and benefits point towards a specific, practical posture for construction businesses.

First, treat AI as a toolbox to be assembled around the dispute, not a single proprietary system to buy into. Off-the-shelf legal research, e-discovery and document analytics tools, properly selected and combined, will comfortably support the great majority of construction disputes.

Second, insist on transparency in how AI-related costs are charged. These should be visible and cost-justified — whether passed through as disbursements or agreed as a fair usage-based fee — rather than folded invisibly into hourly billing.

Third, keep experienced legal and technical judgement firmly in the loop. AI-generated research, document review output, and causal or delay analysis should always be tested and refined by practitioners who understand both the technology’s limitations and the underlying technical and contractual issues in play.

Fourth, treat privilege and data security as a design requirement, not an afterthought. Construction disputes routinely involve commercially sensitive cost data and personal data relating to site personnel, and any AI-assisted review platform or cloud-based data room should sit within a clearly defined privilege framework, with data handling arrangements that reflect the actual jurisdictions involved.

The Role of In-House Counsel

Making good on all of this requires someone inside the business who understands both the legal landscape and the practical toolbox of AI tools now available, and who is positioned to assemble that toolbox before a dispute crystallises rather than scrambling to do so once external counsel has already been instructed.

A senior, embedded in-house function can maintain a pre-vetted panel of AI-enabled research, e-discovery and analytics tools alongside a bench of delay and quantum experts, so that both can be engaged within days of a dispute being identified rather than months later. It can insist on transparent, cost-justified treatment of AI-related spend from external counsel. And it can make the early, evidence-based case assessment that AI now makes possible — feeding a realistic view of merits and quantum into commercial strategy at the outset of a dispute, rather than waiting for a formal expert report many months down the line.

Conclusion

None of the changes described in this article are revolutionary on their own. Greater use of adjudication and mediation, more disciplined case management, and the sensible use of AI as a toolbox rather than a silver bullet are all incremental rather than radical ideas. Together, however, they represent a genuine and overdue response to what arbitration users have been saying for some time: that dispute resolution in the construction sector needs to become more affordable, more efficient, and more accessible, without sacrificing the quality of the analysis or the fairness of the outcome.

This article is published for general information purposes only. It does not constitute legal advice and should not be relied upon as such. C&IDS is a UK limited company (No. 11789861). Steven Hunt is a Solicitor of England & Wales.