ComplyQS
HomeArticles › AI explainability: the five things surveyors must provide on request

AI explainability: the five things surveyors must provide on request

Under RICS 4.4, firms must be able to provide written information about their AI use on request — the system, how it works, the due diligence, risk management and reliability decisions.

A client asks: “this figure — how exactly did the AI arrive at it, and why should I trust it?” Section 4.4 of the RICS AI standard says a firm must be able to answer in writing, on request. Explainability is the exam the rest of your records are revision for.

The five items

On request, RICS-regulated firms must be able to provide written information covering:

  1. the type of AI system used,
  2. its basic ways of working and its limitations,
  3. the due-diligence processes carried out before use,
  4. how the associated risks are identified and managed (your risk register), and
  5. the reliability decisions made about its output.

Why this section quietly matters most

Notice what the five items are: they are the outputs of every other duty in the standard. A firm that kept its systems register, filed its due-diligence answers, ran its risk register and recorded its output reviews can answer a 4.4 request by assembling what already exists. A firm that did those things informally has to reconstruct them under time pressure, for a possibly unhappy client — the difference between retrieval and archaeology.

The realistic trigger isn't curiosity. It's a challenged valuation, a disputed cost plan, a PI insurer's questionnaire, or a complaint. Explainability requests arrive attached to disagreements — which is exactly when you want the answer to be a file, not a scramble. Baseline knowledge (section 2) matters here too: someone in the firm needs to genuinely understand the tools well enough to write item 2 credibly.

Do we need to explain the algorithm itself?

No — the standard asks for the type of system, its basic ways of working and limitations, not source code or model weights. The depth is 'appropriate understanding', consistent with the baseline-knowledge duty in section 2.

Who asks for this in practice?

Clients (the standard frames it as supporting clients who want to understand or challenge outcomes), and in practice also PI insurers and complaint processes. Anyone with a legitimate interest in how the output was produced.

ComplyQS assembles the 4.4 answer from records you already keep — tool details, due diligence, risks and named output reviews, exportable per project.

See a project record — free trial

This article is general information, not legal or professional advice. ComplyQS is not affiliated with or endorsed by RICS.