For Construction Arbitrators and Adjudicators: Will AI Tools Help or Hinder the Decision-Making Process?
Artificial Intelligence (AI) is no longer an abstract concept confined to research labs or the preserve of Silicon Valley start-ups. It is here, evolving daily, and already infiltrating the alternative dispute resolution (ADR) arena. In construction adjudication and arbitration, the question is no longer ‘Will AI have an impact?’ but rather ‘How do we harness it without compromising the fairness, integrity, and enforceability of our decisions?’
For the more seasoned practitioner – and here I speak not only of seniority in years but equally in professional standing – the emergence of AI tools can seem as unsettling as it is intriguing. Those who have long relied on paper bundles, annotated precedents, and face-to-face hearings may view ‘machine learning’, ‘natural language processing’ and ‘large language models’ as alien territory. The learning curve appears steep; the language of the technologists, unfamiliar; and the prospect of ‘getting it wrong’ is a genuine deterrent. However, to disregard the technology altogether would be to risk falling behind the pace and reality of modern practice.
The potential benefits are compelling. AI can ingest thousands of pages of pleadings, witness statements, and exhibits in moments, and then proceed to sort, categorise, and extract the specific passages relevant to a particular issue. It can highlight contradictions in testimony, map timelines of events, and even suggest lines of cross-referencing between the facts in dispute and applicable authorities. Properly deployed, AI can lighten the administrative load, accelerate research, and sharpen the clarity of fact-finding – freeing the arbitrator or adjudicator to concentrate on weighing the evidence and applying their professional judgment to the relevant data.
Yet, the risks are equally real. The Association of Arbitrators recently published an excellent set of AI Guidelines[1] which emphasise principles of accountability, confidentiality, fairness, and transparency. Not all AI tools are created equal or provide equivalent functional capacity. Public consumer-facing platforms such as ChatGPT, DeepSeek, or Perplexity, while impressive in conversational ability, are not designed to handle confidential case material safely. Uploading sensitive pleadings or evidence to an open platform risks a breach of data protection laws, the loss of privilege, and potential prejudice to one or more parties. Furthermore, AI outputs may contain factual inaccuracies – known as ‘hallucinations’ – or reflect hidden algorithmic biases. In ADR, such defects are not mere inconveniences, they can undermine the very enforceability of a decision.
The alternative lies in secure, closed-platform AI systems, hosted either on-premises or within a private cloud, built specifically for legal or arbitral work. These special AI systems can store pleadings, evidence, and transcripts in encrypted form, apply advanced search and analysis, and integrate directly with trusted legal databases, such as SAFLII for South African case law, or international repositories for foreign precedents. Such systems can track your research steps, ensuring that every citation is traceable and every factual statement verifiable.
For more senior arbitrators and adjudicators, the challenge is not to transform overnight into AI technologists, but rather to develop a working familiarity with what AI can and cannot do. That begins with understanding the limits – AI can summarise, categorise, and even suggest patterns, but it cannot decide credibility, assess fairness, or even weigh proportionality. These areas remain the domain and prerogative of human judgment. The wise approach is to embrace AI as a diligent assistant, not as a substitute decision-maker.
The pitfalls are avoidable if we follow structured safeguards: Never rely on AI-generated text without verification, never expose confidential data to insecure systems, and always ensure that ownership of the reasoning process is maintained. Equally, the benefits are substantial: Faster case preparation, clearer organisation of complex evidence, and reduced administrative burden – especially in document-heavy construction disputes.
In short, AI will not replace arbitrators or adjudicators. But those unwilling to engage with it may find themselves overtaken by colleagues who can integrate these tools into their workflow while preserving both procedural fairness and the inevitable human element of decision-making. For the senior practitioner, the path forward is not to resist change, but to lead it – armed with both the wisdom of experience and the prudence of the AI Guidelines.
Chris Binnington
August 2025
[1] Guidelines on the Use of Artificial Intelligence (AI) in Arbitrations and Adjudications (as adopted and approved by the Association’s Board on 15 May 2025).
