Back in the November/December 2023 issue of BIC Magazine, I wrote about the digital revolution in workplace safety and health.
While I’m sure that all our dear readers remember every word of that thought-provoking article (indulge me here), I’ll recap for the new folks:
We’ve come a long way from paper job hazard analysis, pre-task plans and near-miss or incident investigations thanks to technology, which has allowed us as EHS&S professionals to improve training and reporting efficiency while creating more collaborative environments with our front-line workforce.
Or technology is making our jobs easier. Six months have passed since I wrote that proclamation or, in nerd speak, four iPhone iOS updates and one new ChatGPT iteration. Not much has changed in the overall EHS&S scheme of things except for this: more of us are using generative AI to aid in our training and reporting content development.
We’re letting the robots take over.
I’ll be the first to admit that I am a technology Luddite. My nine-year-old daughter knows more about iPad functionality than I ever will. And that’s okay (no, it really is). But I have just enough knowledge of how tech works — and the value it brings to the workplace — to understand that not every change is bad. In this case, using generative AI to help us develop training content is a good thing. To a point.
For the spring issue of VPPPA’s award-winning magazine, The Leader, which focuses on the future of EHS&S, I undertook an experiment for my regular column and used ChatGPT to create an overview of how generative AI can help develop training content for the EHS&S professional.
The end product was, quite frankly, surprisingly good. I simply input key concepts and terms that I wanted to convey, and the generative AI engine produced a cohesive, logical article that conveyed my main points. But … only after I reviewed it.
And therein lies the key to using generative AI as EHS&S professionals: We can input key data points and information into the content engine, and it will scour the corners of the internet to produce cohesive, logical content based on our inputs, but it still needs to be thoroughly reviewed. Need a quick toolbox talk on fall protection? Generative AI can produce it … if your inputs are accurate. Historical data analysis on near misses because of improper tool tieoffs? Depends on the body of inputs from across industries.
The point is this: We should all be using generative AI to create training content and data analysis from our internal, historical incident and near-miss data. The tool itself is far more efficient than our own manual analysis and even some of the static (i.e., non-changeable) platforms that we currently employ. But there’s a catch — and this is where you can relax a bit as AI isn’t taking your job and there’s no need to start learning new skillsets. Generative AI is only as good as the data input into its engine. In other words, if you’re not reviewing the output, you’re putting yourself at risk of not only producing inaccurate training and evaluation content but propagating inaccuracies down the line.
We all know how that turns out if you think about the Swiss cheese model of accident causation, originally proposed by James Reason. It likens human system defenses to a series of slices of randomly holed Swiss cheese arranged vertically and parallel to each other with gaps in between each slice.
Technology is the future — in general, but also in the EHS&S profession. We’ve already adapted to using mobile platforms in the field with great success. Now we’re faced with the opportunity — and challenge — of incorporating generative AI into our training and evaluation processes. The opportunity is to create time efficiencies that allow us as EHS&S professionals to focus on the most critical component of our jobs — sending workers home in the same, or even better, condition than which they arrived onsite. The challenge is to avoid the complacency that technology brings — to stop focusing on continuous safety and health improvement and let the apps and ChatGPT or other tools do our jobs.
Choose wisely, friends.
For more information, visit vpppa.org.
