As AI gets smarter, could it possibly be getting safer? AI assistants and agents are not new to the construction industry, but they are helping construction companies approach, implement and execute their health and safety plans in new ways. The Innovation Solution Center team at Barge Design Solutions sat down with Construction Executive to speak on the evolutions and innovations of health and safety plan assistants in the safety realm of the construction industry.
How prevalent are AI-powered tools becoming in the safety realm of construction?
Like many other aspects of construction, safety is an area that is turning to AI as a solution for improvement. We see a need for producing more and higher-quality safety content, but in a way that is dynamic and reflective of unique aspects in each specific project.
How does safety-focused AI differ from other forms of AI? What further capabilities does AI need for proper safety protocol?
The application of AI in safety scenarios is very similar to other forms of AI, as it relies on the same generative AI principles. However, we tailored the interface to ensure the human user stays in control to ensure that the AI serves to augment the critical safety thinking of our employees rather than replace it.
Does cybersecurity factor into the safety component of this AI?
Yes! Cybersecurity plays two roles with this safety solution. First, the solution is itself secure, in the sense that it complies with all of our security protocols and protects employee and project data. Second, if cybersecurity is a potential hazard when our employees are on-site (e.g., possessing personal or unsecured devices when near critical energy infrastructure), then this solution will help identify that and write protocols to reduce risk. This capability is important as there will be varying safety issues from one project to the next. Generative AI is especially useful in these dynamic scenarios.
What needs to be prepared before the AI tool is capable of writing health and safety plans? Can these plans be tailored per company/per project?
The user needs to enter a description of what employees will be doing on site, and we have structured the interface to ensure this is both easy and flexible for unique project details. The assistant then uses this information to research hazards and safety requirements based on the project’s location and type (e.g., federal or private), synthesize the information it gathered and write a safety plan using a configurable template. This workflow ensures that the output can be structured in a consistent way while adapting to project-specific details. The result is a well-crafted plan, reducing the time to first draft from hours to minutes.
Are there any legal ramifications of safety planning with AI? What needs to be done beforehand to mitigate any potential suits? What types of hazards would this technology present? Especially in terms of liability.
Our employees still control and remain responsible for each safety plan. AI tools are great for automating time-consuming tasks, but at this point we cannot aim for critical workflows like this to be fully autonomous. AI products still need human iteration, thorough review, and regular updates to ensure all outputs meet organization standards and avoid any legal ramifications.
What happens after the drafting process? Does AI help with finalization?
Not currently, as we want to emphasize the importance for human review, but the user can use AI for small revisions while first drafting each section of a health and safety plan. After each section is completed, the user can export the plan to a Word document. It is at the point that the user should review and edit one last time. After that, we like to add each finalized and well-crafted plan to our knowledge base so that the next plan is more informed on the content and structure of a high-quality health and safety plan.
How have companies responded to this technology?
This solution has received significant positive feedback in our company, and several other firms and clients have asked us for help in implementing similar (i.e. writing-related though non-safety) solutions. While the interface and workflow should be recognized, the real star is the generative AI and how we’ve applied it to solve a repetitive, standardized, and mundane task.
What is the prognosis of this tech within the construction safety sector? How do you see if affecting all facets of the larger construction industry?
Our goal is to increase the quantity and quality of safety-focused plans. We set out to make the drafting process easier and slightly entertaining for users, in hope to increase its use and awareness in the industry. However, we also see this leading to new ways to apply generative AI. We think the design of this solution demonstrates how we can move beyond the limitations of the chatbot interface. This same approach of enabling fast information research and synthesis has great potential to improve many workflows across the engineering, architecture, and construction industry.
SEE ALSO: AI AS AUGMENTED INTELLIGENCE: ENHANCING HUMAN CAPABILITIES
