AI is set to change the way we work forever, good or bad; it seems to be here to stay. This is no different in the construction industry. An industry known for sticking to its ways, it is slowly accepting the aid of machine learning to speed up slower processes. We are going paperless, we are moving to construction management software, and soon, we will be making use of AI.
(We have discussed this before when we did a quick dive into Agentic AI, what it is and where you may find it in this industry.)
One space where you may not expect AI to find itself is in construction arbitration, and they’re calling it Alternative Dispute Resolution (ADR). ‘This initiative marks a significant milestone in alternative dispute resolution’ (Aceris Law LLC, 2025). It aims to save time by using AI for document-only arbitration. In disagreements which can often drag on, slowing construction progress and causing significant frustration for both parties, could this be the key? Or is this simply not something an AI is capable of?
We interviewed Dean Dixon, SmartStruct’s founder, on his feelings on the subject. Dean also runs a housing development company and also looks to implement AI into SmartStruct to improve what it can do for his customers.
‘An AI is only as good as the information you give it.’
The immediate answer was, no, it’s simply not ready yet. We don’t know the right questions to ask, and there was clearly a question of trust.
If an AI is expected to make a legal call, it needs the most up-to-date legal information. As AI has been caught out on this in the past, it would need to be as well-informed as its human counterparts. Dean argues it would need to be focused on this task alone to avoid learned opinions that may muddy the water. This is where, for him, Agentic AI comes in.
Speaking on a currently theoretical Agentic AI, or in this case, AI Arbitrator; He indicated that the goal would be to create a ‘mini mediator’ that would live inside SmartStruct. This would be able to collect, read and review documents stored in the SmartStruct project dashboard. This would then be able to consolidate the findings and documents into a clear argument to a human arbitrator.
This is the way Dean envisions this working: there would always be a final human element to allow fair, nuanced and experienced arbitrators to have the final say, and inject that very human missing element to the discussion.
The Holy Grail of resolving construction disputes is causation.
Without this, we run into the black box problem, an inability to clearly see how the AI made its decision. AI with access only to documents will only consider black and white. It would always need safeguards. It is piecing together photos and emails, missing parts of the story, and it can only be as unbiased as its programmers.
Therefore, AI cannot stand alone in these decisions, as it lacks context and nuance. An end-to-end software, including agentic AI, could empower human arbitrators with factual clarity. Providing evidence and information about the day-to-day contributing factors.
In the best-case scenario, it would be less ‘who said what and when’. It would be interpreting the facts within the framework of the law, and each photo, discussion, document and payment logged on SmartStruct would help to build a picture, with context.
For example:
- Communication Logs: Was there a casual Slack message where the client approved a change informally? The AI could find it.
- Design Revisions: Did a late change from the architect cause the delay the contractor is now being blamed for? The AI could correlate the revision timestamp with the work schedule.
- Site Conditions: Did sensor data from the site show a week of unexpected rain, supporting a force majeure claim? The AI could verify this instantly.
- Progress Photos: Could daily drone footage visually prove when work was actually completed, contradicting a dated sign-off?
Certainly, for small disputes, this technology could save precious time and help bring things to a swift conclusion. This then frees human arbitrators up for larger and more complex disputes, which would usually get buried under hundreds of small ones. But it remains important to use this tool as an aid and not become reliant on it, lest it begin to negatively affect human decision-making and judgments. It must always work within a fence of safety and an ethical framework to ensure fair resolutions.