Alex is founder of Voxel, a computer vision AI company. He was previously founder of Sherbit, a healthcare company acquired by Huma in 2018.
getty
I joined the forefront of IoT in 2014 when I built a platform to monitor outpatient vital data using wearable devices. After many trials at major hospitals, I learned a lot about the cutting-edge of health tech. Today, I see a lot of misguided optimism about wearables.
Early smartphones generated a lot of hype about IoT, as everyone rushed to collect data from their devices. AI computing costs were high, and most people weren't thinking about it. But cloud computing has leaped forward in the past decade. Today, we don't need new hardware to generate data—we can take data we once ignored and use AI to extract powerful insights.
In industrial safety, wearables are being used to prevent musculoskeletal injuries. But there's a much richer source of ergonomics data: cameras. CCTV can "see" everything in a worksite: pallets, production lines, cranes, forklifts and other machinery. It's easy to train AI to recognize when a worker is in danger. AI has quickly made safety wearables obsolete.
1. Wearable Data Is Irrelevant
Wearables collect biometric data that's totally irrelevant to safety. If I'm training my team to lift safely, I don't need to know their precise biomechanics. If you're bending while lifting, I don't need to know the curvature of your spine. I just need to know: are you bending? If the answer is yes, I already know you're at risk of injury. Now I need to find out if there's a problem in our environment causing the unsafe lift. And tomorrow, we need to review safe lifting practices as a team, so everyone holds each other accountable.
You can recognize a dangerous lift with the naked eye, and AI learns what it "looks" like the same way. Biometric data is irrelevant to the solution for musculoskeletal injuries: better training and teamwork.
2. Wearable Data Is Incomplete
According to the Bureau of Labor Statistics, musculoskeletal disorders are only one-third of all workplace injuries. What about the rest? Wearables can't detect spills, cranes and other hazards. But AI can "see" these and proactively alert workers to risks.
Some wearables detect "jumps" to alert potential slip-and-fall hazards. But the devices only record a timestamp and location. You have to backtrack to that location before you can diagnose the problem. By then, the scene's changed and the hazard may have disappeared.
When AI sees a "jump," it clips the video and stores a complete visual context. You have immediate visibility into what's putting your workers at risk and you can take it to your team to start a conversation about how to improve.
3. Wearable Data Is Too Expensive
Wearables manufacturers charge a per-user subscription fee to access their data. The costs add up fast. Devices gets damaged in industrial environments, adding maintenance overhead. And they become obsolete after a few years.
Wearables inevitably require extra hardware to function properly. To alert when workers get too close to machinery, you need telematics on every forklift, crane, production line, etc. That means more maintenance costs.
Meanwhile, the cost of AI is fixed to the number of cameras, whether you have 25 workers or 2,500 workers. AI constantly improves in the background with no physical upgrades. Just teach it what an object "looks" like and AI will "see" it—no hardware required.
4. Wearable Data Is Used Incorrectly
If your workers are at risk, what should you do to protect them? Wearables send "haptic feedback," a vibration to alert you're lifting in an unsafe way. Nobody likes being buzzed like a caged rat in a lab experiment. The devices are unpopular and workers must be pressured to wear them.
Safety is a team effort, whether it's two people lifting together or one person steadying a ladder while another climbs it. Engaged workers are more collaborative. They feel a stronger sense of responsibility towards each other. The more your team communicates, the less likely they are to get into an accident. Safety is not an individual problem; it is a collective problem.
Injuries are rarely caused by individual poor judgment. They're usually systemic failures. Somebody didn't get trained properly or something in your process isn't working. The responsibility falls on employers to make sure the workplace is safe. Reducing injuries requires effective training and data-driven adjustments to the work process. Most importantly, it requires a change in the mindset of your team to think proactively about safety.
The Future Of Industrial Safety
Safety culture is a collective mindset. It's how safety is thought about, talked about and actually implemented. Safety culture is usually reactive: Workers are approached after an incident and reprimanded for doing wrong. "Buzzing" workers when they make errors is not the right formula for safety.
The key to reducing injuries is a proactive culture based on communication and accountability. One-third of injuries occur in the first year on the job, and 1 in 8 injuries happens on the first day. Injury prevention requires proper training driven by dialogue and incremental improvement. Prioritize positive reinforcement. Spend more time rewarding good behavior than you do punishing bad behavior. If your team feels like they're risking their job whenever they talk about safety, they won't talk about it at all!
If your team catches near-miss incidents and stops them early, serious workplace injuries could disappear altogether. Thanks to AI, security cameras now offer a superior method of tracking leading indicators of injuries. AI can identify risks, record relevant scenes and recommend adjustments based on a worksite's specific needs. In the future, site managers will regularly use AI to discover "coachable" moments, like a basketball team reviewing the film after a game.
Wearables will inevitably play a role in some industrial operations. They may remain useful in environments without camera access, like underground or underwater. And some types of data (like air quality) can't be "seen" by cameras and require special sensors. But it's only a matter of time before even the most complex work environments are managed by AI. By then, today's smart safety wearables will be a distant memory.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
