TOOLS OF THE TRADE – JUNE 2026
AAA AND THE AUTOMATION OF ADR
Note from the Editor
The Association of Arbitrators (Southern Africa) NPC’s (the Association) AI Committee recently organised a meeting with the American Association of Arbitrators’ (the AAA) AI development team who have already implemented their artificial intelligence (AI) programme, “The AI Arbitrator”.[1] Arising from that interaction, Chris Binnington, AI Committee member and Director of the Association, offers another insight into the fast developing world of AI-assisted dispute resolution.
AI, ADR and the Association
The emergence of the AAA’s AI Arbitrator initiative has intensified a question that only a few years ago seemed theoretical – how close are we to replacing the human adjudicator or arbitrator in construction disputes?
The answer is that we are closer than many practitioners imagined, but still some distance from replacing the human decision-maker entirely.
The AAA’s new AI-assisted arbitration platform represents a significant step forward in the use of AI within dispute resolution. The system is designed primarily for lower-value, document-only construction disputes. It analyses pleadings, organises evidence, identifies issues, prepares chronologies, and can even generate draft awards. Yet the AAA remains careful to emphasise that the final award is still issued by a human arbitrator who reviews and takes responsibility for the outcome.
This distinction is critical.
“The adjudicator or arbitrator must remain personally responsible
for the reasoning process, legal analysis, procedural rulings, and ultimate decision.”
Construction adjudication and arbitration are not merely exercises in document analysis. They involve judgment, procedural fairness, credibility assessments, proportionality, and often a deep understanding of commercial realities and industry practice. An experienced adjudicator does more than process information – he or she evaluates nuance, weighs probabilities, detects tactical omissions, and manages procedural balance between the parties.
This is where the ethical guidance of the Association becomes highly relevant. The Association’s recently published Guidelines on the Use of Artificial Intelligence (AI) in Arbitrations and Adjudications recognises both the usefulness and the dangers of generative AI. The guidelines accept that AI may assist with administrative functions, research, organisation of evidence, language refinement, and drafting support. However, it stresses that the adjudicator or arbitrator must remain personally responsible for the reasoning process, legal analysis, procedural rulings, and ultimate decision.[2]
The Association correctly identifies the central danger – impermissible delegation of the adjudicative function.
If an arbitrator merely signs off an AI-generated award without independently verifying the reasoning, evidence, authorities and conclusions, the integrity of the process is compromised. The problem is not only ethical but potentially jurisdictional. Parties appoint a named adjudicator or arbitrator because they trust that individual’s judgment and expertise, not the output of an algorithm trained on historic decisions.
There are also practical concerns unique to construction disputes. Many cases turn on incomplete records, site realities, witness credibility, programming logic, quantum methodologies, and the interpretation of contemporaneous conduct. AI is extremely capable at identifying patterns in data but it remains weaker in assessing context, commercial behaviour and the “feel” of a dispute which experienced practitioners often develop over decades.
Nevertheless, it would be a mistake to under-estimate how rapidly the technology is advancing. AI systems are already exceptionally effective at summarising submissions, identifying inconsistencies, analysing large quantities of correspondence, and producing first-draft procedural or analytical material. In lower-value disputes, particularly those conducted on documents alone, the economic pressure toward AI-assisted resolution will become increasingly difficult to resist.
The likely future is therefore not the replacement of adjudicators and arbitrators, but rather their transformation.
The adjudicator of the future may resemble less a traditional drafter of decisions and more a judicial supervisor of an AI analytical engine – verifying fairness, testing reasoning, exercising discretion, and ensuring that justice remains human even when much of the analytical work is machine-assisted.
For the foreseeable future, construction dispute resolution will still require human judgment. But the era in which arbitrators and adjudicators worked entirely without AI assistance is already coming to an end.
Director
Missed a Previous Arbitrarily Speaking Issue?
Catch up here.
View the Association’s Code of Ethics.
