Key takeaways:
If you’ve looked into AI cameras, wearables, or predictive sensors for your workplace, you’ve probably heard a lot of promises and not much proof. Every vendor claims their tool reduces injuries, but few have solid evidence to back up those claims.
Dr. Matt Law spends his career closing that gap. He directs the National Safety Council’s Work to Zero initiative, a program that tests safety technology against real-world outcomes instead of taking a sales rep’s word for it. We asked him what AI can and can’t do for injury prevention, which technologies already have evidence of success, and how to tell a real solution from a good pitch.
AI’s role: Decision support, not decision authority
Start with the question most employers want answered: Can AI prevent workplace injuries?
Law’s answer is a straightforward no. Right now, AI can’t directly prevent workplace injuries. It doesn’t replace the judgment calls that keep people safe. What it does is make the people doing that work faster and better informed.
“AI acts as a powerful moderator for the work it takes to prevent workplace injuries,” Law said. It can analyze data faster, automate repetitive tasks, and surface information that supports a decision. It doesn’t make the decision. “It is decision support, not decision authority.”
Law compared it to any other tool on a job site. A power drill gets a job done faster than a hand tool, but only if the person using it knows what they’re doing. AI works the same way. It’s only as effective as the safety program and the people applying it.
“Humans will always be required for the work. Humans are required to make the decisions. Technology will help us get there, but it’s not replacing us.”
Dr. Matt Law, Director, Work to Zero, National Safety Council
Cooling devices and wearables show the clearest results
Which specific technologies actually produce results at this point? Dig into Work to Zero’s own pilot research rather than a vendor’s pitch deck. A recent round of studies on heat stress found real, measurable reductions in heat-related injuries, including one deceptively simple device: A handheld tool that cools blood circulating through the hand, lowering core body temperature.
Wearables built to monitor posture and lifting mechanics have shown similar results, under the right conditions. “When it’s used for its intended purpose, we are seeing measurable improvements on the plant floor,” Law explained.
But implementing a device requires the support of coaching and process changes. Two employers in the same industry can see completely different results from the same tech, depending on how it’s implemented and how the workforce responds to it.
📍 Read next: Smart Safety: How Wearable Tech Cuts Work Comp Claims by 90% >
The two biggest barriers to adoption
One major barrier to adoption is, of course, cost. This is an especially big question mark for small and mid-size employers. Larger organizations can absorb the risk of technology that doesn’t pan out, but smaller companies can’t afford to guess. “Return on investment cost is a huge component,” Law pointed out.
The bigger barrier is human. Employees won’t accept a tool if they don’t understand what data collects and why. “If they don’t know those things, they won’t trust it, and that becomes a huge barrier for bringing in technology into the workplace,” Law said.
Addressing distrust in data collection
This problem isn’t specific to technology. Earning real employee buy-in is a prerequisite for almost any safety initiative. But today, people are wary of any company collecting their data – especially their employer.
That distrust often comes from the fear that the data will be used against them. They might be afraid that they’ll be punished for breaking a rule, or that the data will be used to justify decisions about their role.
Law validated these fears: “You can use it in a punitive way. You can use it to perpetuate what your organization has been doing incorrectly all along.” But if the response to uncovered data is discipline instead of fixing the problem, you’re simply automating a bad system.
Addressing employees’ concerns about data collection takes transparency and communication. Employees need to know upfront what a tool collects, how it will be used, and how it’s protected.
“Organizations have to be absolutely transparent upfront about all of those different components for their employees if they’re going to be effective,” Law summarized.
Vetting a vendor’s claims: 4 questions to ask before you buy
Law fields technology pitches every day. Here are the questions he advises employers to ask before spending a dollar on any new tool.
| Question | What to look for | Red flags |
| How have other organizations used this tech? | Direct references to real customers who have achieved specific outcomes, not just satisfaction. | The vendor can’t provide a single reference, or only offers curated testimonials with no one to contact. |
| Is there independent, third-party research on its efficacy? | Case studies published by an outside group, like the NSC’s Work to Zero program. | The only research comes from the vendor itself. |
| If vendor research is all there is, what does it show? | Transparency about the study’s limitations, funding sources, and methodology. | A white paper or case study touting results without describing methodology. |
| What does ROI look like for my organization, not just what’s advertised? | Independent calculators comparing real-world value to marketing claims. | The vendor discourages you from running your own numbers, or the advertised ROI can’t be verified. |
☑️ The bottom line: A technology with real evidence behind it should be able to survive all four of these questions. If it doesn’t hold up, that’s information too.
Start with one specific problem
The most common mistake Law sees isn’t choosing the wrong technology. It’s starting without a real problem to solve. “You can’t just say, ‘I want us to be safer.’ That’s too big of a problem,” remarked Law.
Instead, name something specific: A recurring strain injury in one department, a heat-related incident pattern during certain shifts, a near-miss trend. Then look for technology built to solve that exact problem. If you don’t find the right fit, don’t try to force something. Sometimes the fix is a process change that costs nothing.
Once you’ve connected a problem to a possible solution, pilot it before committing. Set clear KPIs up front, measure the inputs and the outcomes, and be honest about what the data shows. “Just understand that when you get it into the real world, things might change,” Law said.
Your job as the buyer is to ask good questions before you spend, and continue measuring success once it’s implemented.
Ready to pilot new safety technology? Set clear success metrics before you start. Learn how in our guide: Safety Metrics: How to Measure Your Program’s Success.
Frequently asked questions
Can AI prevent workplace injuries?
Not by itself. AI supports the people who prevent injuries, but it doesn’t replace their judgment. It analyzes data, automates repetitive tasks, and surfaces information faster, but decisions about safety still require human expertise.
What safety technologies have the strongest evidence behind them right now?
Cooling devices for heat stress and wearables that monitor posture and lifting mechanics currently have the most measurable research. Pilot studies from groups like the National Safety Council’s Work to Zero program show positive results from these tools, but they depend heavily on how the technology is implemented and how well it’s matched to a specific risk.
How do I know if a safety technology vendor’s claims are legitimate?
Ask for direct references from real customers and look for independent research rather than relying on the vendor’s own studies. If only vendor-produced material exists, evaluate it with that limitation in mind, and use an ROI calculator to compare advertised value against realistic outcomes for your organization.
Is safety technology worth the investment for a small or mid-size business?
It depends on whether the technology solves a specific, identified problem in your workplace. Smaller employers can’t always absorb the risk of an unproven tool, so piloting on a small scale with clear KPIs before a larger rollout helps prove effectiveness before a larger investment.
How can employers address employee distrust of safety monitoring technology?
Transparency is key. Employees need to know upfront what data is collected, why, how it will be used, and how it’s protected. Employers must commit to using that data to fix problems rather than to punish workers.
How do I choose the right safety technology for my workplace?
Start by identifying one specific, measurable problem, like a recurring strain injury or a heat-related incident pattern, rather than a broad goal like “improve safety.” Then look for technology built specifically to address that problem, and stay open to the possibility that a process change, not a new tool, is the better fix.