“Material impact” in the RICS AI standard: how to decide and record it

Materiality is the switch that turns on most RICS AI standard duties. A practical test for deciding whether AI use is material, who should decide, and exactly what to record.

Almost every duty in the RICS AI standard — the systems register, supplier due diligence, client notices, output reviews — switches on when AI use has a material impact on the delivery of a surveying service. Which makes “is this material?” the most consequential question in the whole framework.

The standard deliberately leaves the judgement with the professional. What it does not leave optional: when a firm determines its AI use will have a material impact, it must record that determination and the reasoning behind it.

A practical test

Ask three questions about the specific use on the specific instruction:

1. Does the output feed the deliverable? AI-extracted quantities in a cost plan: yes. AI tidying your internal meeting notes: no.
2. Would an error plausibly change the advice or the client's position? An AI mis-measuring 8% on a drawing changes a cost plan; a clumsy autocomplete sentence gets caught in review.
3. How much human verification happens between AI output and reliance? Full re-checking by the surveyor pushes toward immaterial; spot-checking or direct reliance pushes toward material.

Two honest observations from firms working through this: the same tool can be material on one instruction and immaterial on another (usage, not branding, decides), and two surveyors in the same office will often disagree — which is precisely why the standard wants the reasoning written down, not just the verdict.

Who should make the call?

The standard requires the record; good governance supplies the who. The workable pattern for most practices: the responsible surveyor on the instruction makes the call (they know the usage), against criteria set firm-wide in the AI policy, with a named senior owner for edge cases. Avoid both extremes — a committee for every decision doesn't scale, and “whoever installed the tool decides” isn't governance.

What the record should contain

• The instruction/project it relates to, and the AI system in question.
• The decision — material or not — and the reasoning, referencing the actual usage.
• Who decided, and when.
• What the decision triggered: register entry, client notice, output-review plan.

A decision with a name and a date is evidence; an unattributed “immaterial” in a spreadsheet cell is an argument waiting to happen. For the walkthrough of recording this in ComplyQS, see the materiality decision guide.

Does 'material' mean 'important' or 'big'?

Neither exactly — the test is impact on delivery of the service. A small AI step in a critical calculation can be material; heavy AI use on something incidental to the service may not be. Judge the specific use on the specific instruction.

Do immaterial decisions need recording too?

The standard's explicit record duty attaches to material determinations. Recording the immaterial calls as well is cheap insurance: it shows the question was asked, and it is the natural place to capture the reasoning if usage later changes.

This article is general information, not legal or professional advice. The materiality judgement is always the responsible professional's own. ComplyQS is not affiliated with or endorsed by RICS.