AI begins to change how Hong Kong builds

Government projects have already reached full BIM adoption, while take-up among developers, main contractors and subcontractors has continued to rise. AI, he added, is following a similar trajectory towards routine application.

“The technology is already there and is becoming more mature. What is holding some people back is leadership understanding, uncertainty about returns and whether teams are ready to work differently.”

Recent government policies requiring the use of AI-enabled tools, robotics, drones and remote-controlled equipment on public works contracts marked a change in direction, said Ho, adding that the requirements would influence practices across the broader market, including private-sector works.

CIC is supporting the transition through executive training, funding schemes and shared platforms, with a target of introducing more practical AI use cases this year. These cover areas including site safety, procurement, programme planning, quality control, sustainability and robotics.

A global AI and smart construction conference that includes an exhibition and AI award presentation is also scheduled for June this year, which will bring overseas solutions and suppliers to Hong Kong.

“The focus this year will be on helping firms advance from early experimentation, particularly small and medium-sized contractors that often lack the resources to assess new tools independently,” said Ho.

Access to funding and shared platforms is key to that effort, with existing innovation schemes having helped but not yet reaching enough of the supply chain.

Closer collaboration with tech providers is also needed to shorten adoption cycles and reduce risk for smaller firms, with public funding better used to lower entry barriers rather than support one-off trials.

Government policy is increasingly aligned with that approach, with requirements on robotics, digital tools and AI-supported systems creating a clearer business case for investment.

“The next stage will be to align those requirements more closely with skills development, procurement practices and project governance, so that technology use translates into tangible improvements on site rather than additional layers of reporting,” said Ho.

The call for organisational change was echoed by industry speakers who cautioned against treating AI as a layer added to existing practices.

YY Yip, deputy director of the design management department at Henderson Land Development, said the developer had been exploring artificial intelligence, digital twins and computer vision for several years, but had also paused and reassessed some initiatives when outcomes proved difficult to measure.

Henderson Land has reduced costly design changes on large private-sector projects through tighter design control, standardisation and BIM-led workflows, driven by process discipline rather than automation alone.

Technology would only deliver value, said Yip, if workflows were revised and staff were expected to work differently, instead of treating new systems as tools layered on top of established methods.

From a tech firm’s perspective, Fred Sheu, national technology officer at Microsoft Hong Kong, said companies were shifting towards a working model in which staff manage AI “agents” that handle routine tasks, allowing people to focus on judgement, co-ordination and oversight.

He said AI was beginning to alter job roles rather than eliminate them, with employees increasingly expected to direct and review automated processes. Organisations that delayed adoption, he reminded, risked losing ground to peers that treated AI as part of core operations rather than an experimental add-on.

Will Cavendish, global digital services leader at Arup, said AI adoption across architecture, engineering and construction had progressed faster than earlier digital tools, but productivity gains at organisational level took time to materialise.

He said AI delivered the strongest results when combined with domain expertise and revised working practices. In design and engineering work, bespoke systems trained on project and engineering data were often required, rather than generic platforms.

Steve Lewis, partner & head of infrastructure at EY, said fragmented data remained a major constraint across construction and infrastructure delivery, inevitably slowing decision-making and limiting the use of advanced tools.

He said AI could be used earlier to organise and connect data sets, so organisations can work with incomplete or unstructured information and there is no need to wait for fully standardised data before deploying new systems.

Several other tech start-ups also presented applied AI tools already in use on construction and infrastructure projects.

Their demonstrations included camera-based systems for site safety and equipment monitoring, drone-enabled progress tracking integrated with BIM models, and predictive platforms designed to flag cost and programme risks earlier in the construction process.

Other solutions focused on integrating data across design, procurement and site operations to reduce manual reporting and support day-to-day decision-making.