More data won’t mean fewer arguments – and judgement, not AI, will decide who wins them, writes Nicholas Viljoen from law firm HFW.
Artificial intelligence is much discussed, usually in abstract terms, but that misses the more interesting point that AI is already having a concrete, practical effect on how work gets done – and nowhere more so than on construction projects and the disputes they generate.
“Reality capture” technology now makes it possible to build a richly detailed visual record of a project from start to finish. Those records will change how party-appointed experts handle evidence and form their opinions, and those same expert reports are now subjected to forensic analysis by lawyers using AI to leave no digital stone unturned.
Starting with project execution, reality capture is really a family of technologies – drones, lidar scanning, 360° cameras and cameras mounted on personnel walking the site. What is new is not the recording, but the intelligence layered onto it. AI is turning reality capture from a passive method of data collection into something closer to an automated site-intelligence system, in which the raw photographs, video and logs that once required laborious human review are analysed automatically and presented as developed findings.
Not every headline example is strictly AI. For example, some reality capture technology enables overlaying an unbuilt model of electrical services onto live footage of the as-built works. That is not AI, it is really augmented reality, but the endpoint is more or less the same – an organised, measurable, three-dimensional record of the works at successive points in time.
This matters for anyone involved in construction disputes because the completeness of the project record has always been fundamental. Take forensic delay analysis for example, a recurring source of disagreement between opposing delay experts is typically the as-built record itself, and that disagreement is usually driven by gaps or ambiguities in the contemporaneous documents.
Where a contractor has sent personnel on daily site walks with 360° cameras, those gaps should be fewer and agreement on the as-built should come more readily. A word of caution however. More data does not necessarily guarantee more agreement. To the contrary, it could simply move the argument elsewhere. That is, to what the camera did not capture or to the interpretation of what it did.
Recording everything has consequences on both sides of a dispute
The ability to capture the progress of the works in extraordinary detail may not always be an advantage to one or the other party. A continuous, time-stamped record favours whichever party the facts happen to help and may cut both ways. The same daily footage that might prove a contractor is entitled to an extension of time may, at the same time, record defective work, unsafe practices and so on.
Comprehensive capture expands the body of disclosable material and creates contemporaneous evidence that cannot later be magicked away. Taking a partisan view, a decision to “record everything” may not be an unambiguous good – it is a strategic decision with consequences on both sides of any future dispute.
On the lawyer’s side of the table, AI has been transformational. The specialised legal AI tools now widely available accelerate work that used to be slow and expensive – document indexing, chronologies, and the analysis of the vast as-built datasets that construction projects are notorious for generating.
We can now interrogate, at scale, whether the description of a document in an expert report is a faithful account of the underlying record – a task that manual review made impractical, if not uneconomic. But the technology that interrogates also fabricates. Judges have already had to grapple with fictitious authorities, generated by AI, that found their way into material put before the court. Verification, in other words, is no longer a clerical afterthought, it is the necessary discipline on which everything else that lawyers do rests.
Where does this leave us? If AI makes claims cheaper and quicker to produce, there will be more of them, and they will arrive looking better organised and more persuasive than project claims traditionally have. However, the same technology that builds a claim can take it apart. The contest does not disappear; it moves. It may be that the scarce commodity in the disputes of the near future will not be the capacity to assemble a record or marshal a claim – AI will do both, at volume.
In fact, the scarce commodity may well be the very human quality of judgment – knowing what actually matters, and which immaculately presented claim is hollow underneath. That, for now at least, remains a human task.
Nicholas Viljoen is an associate at the global law firm HFW.
