Responsible AI and human review

Architecture Social may use artificial intelligence and other automated tools to help organise information, draft or summarise content, support search and answer website questions.

These tools support people. They do not replace the judgement, responsibility or direct communication expected in recruitment and client work.

A person remains responsible for every decision about a candidate, client or member of the public. Architecture Social does not use AI alone to screen, rank, shortlist, reject or select candidates.

AI-generated suggestions, summaries or matches must be reviewed by a person before they influence a recruitment action or external communication.

AI may assist with limited tasks such as:

• summarising information already provided to Architecture Social;
• helping draft content or routine correspondence for human review;
• supporting search, classification or the organisation of records;
• suggesting points for a recruiter or client to consider; and
• providing ArchGPT answers that a visitor can check or discuss with the team.

An output is not treated as fact simply because a system produced it. Important names, claims, qualifications, decisions and messages must be checked against the underlying information.

Where AI is materially involved in a service or communication, Architecture Social aims to explain that involvement in clear language. Visitors can ask for a person to review an answer or discuss a recruitment matter directly.

ArchGPT is an AI-assisted website service and can make mistakes. Its answers are not a substitute for professional, legal, financial or employment advice. The website terms explain the expectations that apply when using AI-assisted tools.

Personal and confidential information must be handled according to the privacy notice. Staff should use approved systems, limit information to what is needed and avoid entering sensitive material where a tool is not approved for that purpose.

Human review includes checking for unfair assumptions, missing context and outcomes that could disadvantage a person. Protected characteristics must not be inferred or used as a shortcut for assessing suitability.

If you think AI or another technology has affected you unfairly, email privacy@architecturesocial.com or use the complaints process. Architecture Social will arrange human review of the concern.

This public statement reflects AI and Technology Use Policy v2.0, approved on 28 July 2026 and effective from 1 August 2026. It is scheduled for review by 1 February 2027 or earlier if the service or applicable requirements materially change.