This interview has been edited for length and clarity.
Artificial intelligence (AI) is changing the way many industries do business. Matt Law, DrPh, CSP, REHS, manager of customer safety strategy at Grainger, joined “The Case for Safety Podcast” to speak with host Scott Fowler about how AI is impacting occupational safety and health (OSH) and how safety professionals can use these tools to identify hazards and improve risk management.
Fowler: What is artificial intelligence?
Law: There's a lot to learn about AI, but a good way to define artificial intelligence is as a computer system that shows behavior in a way that someone could interpret as human intelligence, provided the system is dealing with pattern matching.
The key question to ask of anything that could be AI or claims to be AI is whether it is really intelligent or just designed to seem intelligent. A system may appear intelligent when it is really just matching patterns based on the information provided to it.
The reality is that most AI that we see today is still considered very weak. Truly strong AI is still very much science fiction. Machine learning is the most prominent and promising form of AI we see today. This is a process in which machines start by identifying patterns and then learn more as they work through additional data. These systems use data like we use our five senses and continuously grow or learn through observation.
Fowler: What are some practical applications of AI you’re seeing used in workplace safety?
Law: Right now, the only true version of AI that's actively being marketed is computer vision. This is a process where a system analyzes images and video captured during safety inspections or monitoring processes. Organizations can use these systems to identify hazards, equipment malfunctions and unsafe practices.
Computer vision is extremely useful for proactively identifying issues that humans would normally miss. Other potential applications of AI include data analysis, voice and pattern recognition, risk prediction, early warning systems and natural language processing for compliance monitoring.
Some existing safety and health software firms are piloting the augmentation of their systems using AI. But we haven't really seen a true form of AI marketed for these applications.
Fowler: What are the implications of AI technology on the future of safety?
Law: It's hard to predict. Realistically, AI has only recently gone mainstream insofar as people have heard of it and are asking questions about its use. Therefore, we can't really put a number to how we think AI application in safety will grow in the future. However, there is a lot going on in the market and there's more demand.
Personally, I think this will look similar to OSH software trends in the last decade or so. The software market for tracking inspections, incidents, safety data sheets and other things in safety has grown tremendously.
We have more firms offering safety software than ever before. That market is starting to plateau, but it continues to grow as it becomes more affordable and accessible to those who need something better than pen and paper and spreadsheets to get their work done. Even though right now there are only a handful of firms offering true AI solutions in safety, I think this will grow exponentially as the prospective applications I mentioned become real and organizations start to discover the value of implementation.
Fowler: How would you encourage safety professionals to use this technology?
Law: The first thing I’ll say is that there's no need for safety professionals to start learning how to program. I don't build software for safety, but I buy software. Right now, it's more important for safety professionals to get a baseline knowledge of what AI is and what it could potentially do in safety.
Safety professionals need to start brainstorming about issues they think could be solved by AI and start pushing on the vendors who develop AI to solve these issues. As AI continues to evolve, safety professionals will be able to use it to augment their work, solve problems we couldn't solve before and start to make a significant impact on workplace safety.
AI and machine learning cannot, should not and really are not intended to replace the human. They are intended to enhance the human’s ability to perform.
Fowler: How can safety professionals familiarize themselves with this technology?
Law: There are a lot of courses out there on AI, and I think safety professionals should use those to build foundational knowledge and start brainstorming about how it can and will be used in safety in the future. Safety professionals should start asking questions of the vendors that are currently marketing computer vision in safety and the ones that are starting to use AI to augment their software platforms.
Ask questions about the effectiveness, the implementation process, data privacy and the ethics behind AI. These are all things that are going to matter as this continues to develop and evolve.
Fowler: Are there ways AI could make certain tasks easier or processes more efficient?
Law: Safety professionals spend a lot of time gathering data, and often it just sits there. We're not able to put it together in a way that helps us address issues. I think what's going to happen with AI is that we're going to have a better, more efficient way to analyze data.
As I mentioned before, AI is demonstrating intelligence, it's not actually intelligent. It's not able to process things in the same ways that we do. When we gather information, if we gather poor information, we're going to get poor outcomes from that data output from AI.
We have to think about how we gather data, gather the right data and look for the right outcomes when we use that data. The only way we're going to train AI to be useful is if we're training it in the right way. Safety professionals have to make sure they’re looking at safety in the right way and feeding AI the right data.
Think about how difficult it is for a safety professional to explain what they do to somebody who's not a safety professional. In the same way, we’re going to have to have that conversation with AI to make sure that AI is able to help safety professionals and do its job well.
Another consideration is data privacy. We've had a lot of different things happen with data privacy over the years with software in general — data leaks, privacy breaches, whatever it is. You have to protect the data you have and use data encryption systems. You also have to be transparent with your workforce about the data you’re gathering, how you’re using it and what it means to them and the workplace.
Fowler: Anything else you’d like to add?
Law: As a researcher, I’ve found that we still don't have a lot of information about current AI application and safety. If you have an example of how you have used AI in safety, I'd encourage you to please reach out to me. I would love to collaborate and and hear your story.
Understanding Risk Management and Assessment
Get the resources and expert guidance you need to improve how you assess risks to prevent hazards, protect workers and safeguard equipment.
Learn more