Artificial intelligence is changing how construction claims are prepared. But faster submissions should never come at the expense of accuracy.

I believe AI has an important role to play in construction dispute resolution.

Anyone who has prepared or responded to an adjudication knows how demanding the process can be. Correspondence, technical reports, payment applications and contractual documents must often be reviewed within tight deadlines.

A statutory adjudication decision is ordinarily due within 28 days of referral, although extensions are possible in certain circumstances.

AI presents a genuine opportunity to improve how construction claims are prepared. But the responsibility for getting the evidence right must remain with the professionals relying on it.

Where AI can make a difference

Consider a contractor preparing a payment claim involving hundreds of variations, conflicting instructions and months of correspondence.

Building a chronology and identifying the relevant documents can take considerable time.

AI tools can help organise correspondence, identify inconsistencies, summarise technical information and highlight issues requiring further investigation.

Used properly, they allow construction professionals and legal advisers to spend less time processing information and more time examining the issues that determine entitlement.

But processing information faster does not necessarily mean getting the answer right.

When convincing information is wrong

Generative AI can produce information that looks credible but is inaccurate. This includes non-existent legal judgments, incorrect contractual interpretations and summaries that misrepresent the underlying documents.

Imagine an AI tool reviewing months of project correspondence. If it overlooks a revised instruction or treats an earlier version as the final agreement, the resulting claim could present a misleading picture of events.

The summary might read well, but it does not accurately reflect the evidence.

In adjudication, such errors can undermine a submission’s credibility and make the dispute harder to resolve.

The problem is not the technology itself. It is treating unverified output as established evidence.

The Bar Council’s November 2025 guidance on generative AI recognises its potential while emphasising accuracy, confidentiality and professional responsibility.

The industry should develop effective verification procedures rather than dismissing technology that could improve dispute preparation.

Confidentiality and professional judgement still matter

Construction disputes often involve sensitive commercial information, including tender prices, subcontractor agreements, payment records and internal communications.

Uploading these documents into an AI platform without understanding its data handling, security and retention arrangements can create unnecessary exposure.

AI Is not automatically unsuitable for confidential material, but the chosen system must provide appropriate safeguards and comply with relevant contractual, professional and data protection obligations.

Additionally, for me, AI should support professional judgement, not replace it.

Three checks should be non-negotiable.

First, verify every legal authority. Check citations against reliable legal sources and read the actual judgment.

Second, trace factual assertions back to the original evidence. A summary is no substitute for the underlying instruction, payment notice or site record.

Third, establish responsibility for the final submission. Someone with appropriate expertise must review the arguments and approve the material being presented. AI cannot take responsibility for the final submission.

These checks make AI’s benefits more reliable and commercially useful.

Better claims, not just faster claims

I do not believe the construction industry benefits from resisting technology simply because established methods feel familiar.

AI can help professionals navigate large volumes of information and focus their attention where it matters most. But interpreting contractual obligations and understanding a project’s commercial realities still require professional expertise.

The objective should never be to produce the fastest construction claim. It should be to produce a well-supported claim more efficiently.

My position is that AI-assisted adjudication submissions should become an accepted part of professional practice, provided the evidence is independently verified and responsibility remains clear.

Should AI-assisted adjudication submissions face stricter verification requirements than traditionally prepared claims? Or should the same professional standards apply to both?