As AI reshapes construction sites, safety-focused leaders and their teams will need to start deployment plans and continuous training to unlock the technology’s full potential and avoid risks. By Cameron Cooper
Maureen Hassall acknowledges that the jury is still out on the ultimate impact that artificial intelligence (AI) and large language model programs will have on the construction industry, but she is confident of one thing – they have the potential to save lives, if used wisely.
“I firmly believe that from a safety perspective we can get AI to help us prevent fatalities,” says Professor Hassall, director of the Sustainable Minerals Institute’s Industrial Safety and Health Centre at the University of Queensland. “If you look at the fatalities in construction, they’re often around vehicles, working-at-heights types of scenarios and electricity. With AI, those fatalities are absolutely preventable.”
Professor Hassall notes that AI is delivering life-saving safety information in practical ways. The Australian Institute of Health and Safety, for instance, has an AI-powered tool known as SPARK that gives health and safety practitioners quick access to critical information and guidance on state-based legislation and regulations.
“And it’s already writing workplace health and safety (WHS) procedures,” she says.
In the construction sector and beyond, there is a view that AI and digitalisation can create safer job sites, with advanced analytics and machine learning identifying worker-injury patterns, as well as site and equipment hazards. AI-enabled sensors, wearable devices and computer-vision systems can also use real-time data to track worker movements and flag risks such as proximity to moving plant, excessive heat exposure, or unsafe worker behaviours.
However, Pocketknife Group safety transformation coach Cam Stevens warns that AI is not a panacea. While AI can streamline safety processes that are overly bureaucratic and administrative, he says the foundational safety management system processes specifically, quality investigations, robust risk assessments and effective leadership practices supported by trust and culture – are often weak or undefined, limiting AI’s usefulness.
Stevens’ message? “AI benefits those with established, quality systems; in weaker organisations, it serves to amplify underlying issues,” he says.
He encourages construction-site managers to take a holistic view whereby AI contributes to well planned, well-communicated and well-understood work activity, including through supply chains and the sequencing of trades. “Safety comes from better planning.”
DEALING WITH DATA
As construction firms face increasing pressure to deliver projects faster and more safely, the integration of intelligent systems is emerging as a critical enabler of safer worksites. Stevens says the most well established application of AI for WHS is computer vision, a subfield of AI that allows computers to understand and interpret visual information from images and videos. The health and safety applications for computer vision include purpose-built hardware and software systems that provide real-time alerts for high-risk plant and worker interactions. Think, for example, of cameras on heavy equipment that produce large volumes of data on unsafe acts and conditions.
However, Stevens says many sites lack the capacity or culture to respond effectively to this influx of data. “They’re often overwhelmed by what to do with the data, or respond punitively rather than with curiosity”
Stevens also urges caution about the marketing hype that surrounds some predictive-analysis technology that can, in theory, alert site managers to elevated risk information. In his view, some of these products and solutions are misleading in their capabilities and “can cause a lot of confusion in the market around what is and what isn’t possible” with AI and other technologies.
Some experts also have concerns about the psychosocial risks associated with the use of AI in workplaces, especially in relation to employee privacy and surveillance. This has led to calls for companies to implement transparent data-governance frameworks that help ensure information and data is collected, stored and used ethically.
PSYCHOLOGICAL SAFETY
The Australian Psychological Society, for one, is worried about the potential negative impacts of the widespread uptake of AI, suggesting that “without strong safeguards, the rapid spread of AI across workplaces risks a loss of staff trust, increasing work health safety breaches, and worsening mental health outcomes”.
The society calls on businesses that are integrating AI tools to “match their investments in technology with careful planning, robust training and attention to the human factors that drive sustainable performance”. The key is to make sure people feel secure, informed and valued as their workplace adapts.
To combat potential psychosocial and mental health risks, Professor Hassall recommends properly informed risk assessments prior to AI implementation because “it’s a technology that can really be helpful, and it’s a technology that can also become very harmful.”
SETTING A CLEAR VISION
Stevens agrees that strategic, holistic planning is crucial for leaders who are grappling with the safe deployment of AI. “You need to be able to turn insight to action,” he says.
For example, Stevens notes that machine-learning algorithms can be applied across tools such as CCTV or drone camera feeds to understand hazard-risk interactions on job sites. Likewise, putting computer vision systems under a crane will typically uncover numerous examples of critical risk interactions occurring in site exclusion zones.
They key, though, is to use the data to design better, safer work by asking open questions about why such risk interactions are occurring. “Sustainable change is more around culture than it is around the technology,” Stevens says.
The rapid pace of technological change also necessitates new workplace policies and skillsets. Employees need training to understand AI-supported tools, interpret digital alerts and collaborate effectively with automation.
Professor Hassall says leaders must “do the analysis of the risks”, including through subject matter experts and consultation with workers. “We have to treat AI as a black-box technology that we test and prototype before we launch it at large scale.”
WORKING AS A TEAM
With AI poised to deliver major safety gains for construction workers, the challenge will be to balance innovation with trust, transparency and strong governance – in essence, creating worksites that are not only more efficient, but profoundly safer. In this environment, Stevens believes robust, human-centered change management strategies will be required to successfully – and safely – roll out tech solutions. Similarly, leaders and workers will require ongoing education and learning to understand how AI helps rather than threatens them.
Professor Hassall adds that WHS professionals who understand how to develop, critique and leverage AI will be in demand. However, all members of organisations must step up.
“That’s one thing I’ve learned working in AI – it takes the technical development-type people working with the subject matter experts and the end users. It’s going to take teamwork to get this done.”
Photography: XTREKX – 123RF
