AI Can Amplify Expertise. It Can't Replace It.
Why transparency about AI use matters more than hiding it in safety leadership.
Let me start by saying I use AI. A lot. I use it for research, training ideas, writing, brainstorming, and sometimes just to help me get an idea out of my head and into something that makes sense to someone other than me. I think it’s an incredible tool, and I’m pretty open about using it.
In fact, Safety Specific is built around it. AI is integrated into our safety platform, we have Safety AI available as a resource for safety questions and support, and we use AI to create training videos that can take situations people actually encounter in the field and turn them into visual learning experiences.
So, this definitely isn’t an article about why people shouldn’t use AI. Quite the opposite. What got me thinking about this was an interaction I had recently with another safety professional. He sent me a toolbox meeting he had “designed” that morning. I looked at it and immediately thought, ChatGPT.
Not because it was bad. It wasn’t. But if you use AI regularly, you start recognizing certain things—the formatting, the wording, the icons, the way information is organized. It was all there. And I really didn’t care that he used AI. What struck me was the need to present it as something he had designed himself. Why?
There’s nothing wrong with saying, “I had an idea for a toolbox talk, used AI to help me build it, and here’s what I came up with.” That still requires an idea. It still requires knowing what you want. And hopefully, it requires enough knowledge to look at what AI gives you and decide whether it’s actually any good.
AI can write a toolbox talk that looks great. It can help draft a procedure, build a training outline, suggest corrective actions, or give you twenty ideas you hadn’t considered. But it doesn't know your jobsite unless you tell it. It doesn't know that the piece of equipment mentioned in the procedure isn't actually available. It doesn't see that the crew has found a workaround. It doesn't know production is three hours behind and people are starting to rush. And it definitely doesn't carry the accountability if the advice is wrong.
You do.
That’s exactly why we’ve approached AI the way we have at Safety Specific. The goal was never to create technology that replaces the safety professional. It was to give the safety professional better tools.
Our platform brings project- and field-level risk assessments, inspections and audits, training management, safety KPIs, project planning, and other safety functions into one place. Safety AI adds another layer by giving users an AI resource specifically within their safety workflow. And our AI-powered training videos allow us to create realistic, job-specific scenarios instead of relying entirely on generic training content. To me, that’s where AI gets exciting. Not replacing experience. Giving experienced people the ability to do more with it.
The ability to generate information is becoming incredibly easy. Knowing what to do with that information isn't. And that brings me to something else I’ve noticed in our profession that has nothing to do with AI. Sometimes we’re terrible at just talking to each other. Someone brings up a situation because they genuinely want another perspective, and within a few comments, somebody is quoting a regulation everyone already knows. Someone else is correcting a detail that has nothing to do with the actual question. Then somebody comes along to explain what obviously should have happened after everyone already knows how it turned out.
At some point, it stops being a discussion and starts becoming a competition over who can sound the smartest. I don't understand what we gain from that.
I’ve been doing this for more than 25 years. I’ve worked in heavy civil construction, infrastructure, ports, marine terminals, and chemical operations. I've learned a lot.
I've also had plenty of moments when another person looked at something completely differently than I did. Sometimes that person had less experience than me. Sometimes they weren't even a safety professional. That doesn't make their observation less valuable.
A craft employee who performs the work every day may see something I don't. A safety professional from manufacturing may approach a problem differently than I would coming from construction. Someone five years into their career may have dealt with a situation I've simply never encountered. That's the whole point of having peers. We're supposed to learn from each other.
AI should make that easier. We have access to an incredible amount of information and a tool that can help us research, organize ideas, question assumptions, develop training, and communicate better. Use it. I do. Safety Specific does. But put your own experience into what comes back. Question it. Tell it when it's wrong. Ask it to try again. Take the answer into the field and see whether it actually makes sense.
And if AI helped you create something, just say so. There's no shame in using a good tool. I'd much rather have a conversation with someone who says, “I used AI to help me work through this, but I'm not sure I agree with this part. What do you think?” than someone trying to convince the room they already have all the answers.
That's also the kind of community I want Safety Specific to become. A place where technology and real-world experience can work together, and where safety professionals can bring the weird situation, the question they aren't sure how to answer, or an idea they want another set of eyes on.
Tell us what you tried. Tell us what didn't work. Disagree. Change your mind. Maybe even admit you don't know. That's where I think we actually get better at this. AI can amplify expertise. It can help us do some pretty incredible things. But it can't give us years in the field, and it can't take accountability for what we decide to do with its answer.
That part still belongs to us.
Safety Specific: Compliance into Confidence.
