Lifeguard & Ski Patrol

Lifeguard/Ski Patrol scores 72/100 on this site's structural exposure measure, and the real-world deployment picture backs that up. AI-powered drowning-detection cameras like Lynxight (over 1,000 pools across 16 countries) and MYLO are genuinely widespread now, and ski resorts including Park City Mountain and Solitude have installed remote-triggered avalanche towers (Wyssen RACS) so patrollers no longer have to ski into a slide zone to throw a hand charge.

But every one of these deployments is explicitly built and marketed to support the humans on duty, not replace them: Lynxight's own materials describe the technology as making "every guard on deck more capable," and it's being adopted specifically because the industry is short on lifeguards, not because it has too many.

The US, UK, Germany, and France are all reporting lifeguard shortages severe enough to close pools, and Germany logged 411 drowning deaths in 2024 tied partly to reduced pool access, the opposite of a labor surplus story. Most US states also legally require a certified lifeguard, personally holding a current Red Cross, YMCA, or equivalent certification, to be physically present whenever a public pool is open (see, for example, Maryland's COMAR 10. 17. 01.

40 and California's Title 22 Section 65539), which keeps a human decision-maker in the loop by law, not just by convention. A dedicated search for a fully autonomous rescue robot or drone actually replacing on-duty lifeguards or ski patrollers, rather than assisting them, turned up none operating at commercial scale.

The detection-versus-response distinction matters more than the headline AI-camera numbers suggest. A drowning victim typically has a window of only a few minutes before a rescue becomes a recovery, so a computer-vision system that flags a submerged swimmer a few seconds faster than a human scanning the same water still needs a physically present lifeguard to leave the stand, swim out, and extract the person in time, the camera changes when the alarm sounds, not who answers it.

The same logic applies even more narrowly to avalanche-control automation: a remote-triggered tower like Wyssen RACS replaces one specific, dangerous task, manually skiing into a slide zone to place an explosive charge, but it does nothing for the rest of avalanche work patrol still owns, assessing which slopes need control in the first place, searching for and extracting a buried victim if a slide does happen, and administering trauma care on-site, all of which remain entirely human, hands-on work today, and none of it shows up in marketing material for the tower itself, which sells the explosive-trigger step alone rather than the full avalanche-response job.