As artificial intelligence becomes embedded in construction, the sector faces a critical challenge: how to harness its potential while keeping people, accountability and professional judgement at the centre of decision-making.
- AI offers major opportunities to improve productivity, safety and decision-making across construction, but adoption is moving faster than regulation and sector guidance.
- Clear governance is essential to manage risks around accountability, liability, data privacy, bias, transparency, intellectual property and workforce impacts.
- AI should support human judgement, with robust review, assurance and explainability built into how tools are used.
The CE Digital Group session on AI and ethics brought together members of the Constructing Excellence community to explore how artificial intelligence is already affecting construction, and how the sector can adopt it responsibly. Misbah Rafique, a PhD researcher in machine learning, data science and AI at the University of Hull, opened the session by framing AI as a tool with significant potential to improve productivity, safety and decision-making, provided it is used with clear ethical controls.
A key theme was the gap between rapid AI adoption and slower progress on regulation, standards and sector guidance. Participants described the current environment as a “gold rush” or “Wild West”, with organisations testing new tools before robust guardrails are in place. Risks discussed included accountability, liability, data privacy, intellectual property, bias, transparency, safety, security and impacts on jobs and skills.
Examples included AI-generated tender responses, site safety monitoring, computer vision for PPE compliance, drone footage, robotic automation and AI-generated project documents. The group noted that AI may speed up work while making it harder to trace who made, checked or approved a decision. In construction, responsibility cannot be delegated to a machine: any output issued by a professional or business must still be reviewed, evidenced and governed through appropriate quality processes.
Trust and transparency generated detailed debate. Some argued that transparency is needed because AI cannot yet be trusted; others noted that trust is built through repeated evidence of reliable performance. Participants drew parallels with established assurance processes, including check-review-approve systems, ISO 9001 quality management and audit cycles where oversight may reduce as confidence grows, but never disappear. One clear conclusion was that trust may earn less oversight, but never zero explainability.
Used well, AI could help construction improve productivity, develop new materials, support safer sites, target interventions more effectively, reduce waste and deliver schools, hospitals and infrastructure faster. Realising these benefits will require clear governance, workforce upskilling and a human-centred approach.
The session concluded that AI should support human decision-making, not replace professional judgement. Constructing Excellence will continue exploring AI through future sessions on research, procurement, health, safety and wellbeing, while signposting trusted resources to help the sector innovate responsibly.
