■ SAFE RISK ■ Public Service & Government
Not soon, though this is the rare low-risk job where the machines are genuinely trying. Remote vehicles now do most of the searching, but the final approach to a corroded, silted, unpredictable munition still ends with a human on a rope.
“Diving to disarm underwater ordnance. Robots help. But someone still goes in the water.”
Our AI replacement risk score — how we score jobs
A clearance dive is the end of a long process. Survey vessels run side-scan sonar and magnetometers over a grid, contacts are logged and prioritised, and only then does a diver enter the water, often in near-zero visibility, working by touch on a search line. The diver identifies the object, distinguishes ordnance from scrap, assesses fusing and condition, and either places a countercharge for demolition in place or, rarely, renders it safe and recovers it. Around the dive sits an enormous safety apparatus: gas management, decompression planning, tender communications, exclusion zones and shipping notices.
Automation has taken over the front half convincingly. Autonomous underwater vehicles now run search patterns for days, classification models sort sonar contacts and cut false positives dramatically, and remotely operated vehicles carry cameras and manipulators to inspect and often to place charges without a diver in the water. Navies and commercial UXO contractors clearing wind farm sites deliberately design for unmanned intervention, because the cheapest casualty is the one that never happens. On open, deep, well-mapped seabed, diverless clearance is already the default.
The residual is stubborn. Harbour mouths, river deltas, surf zones and heavily silted historical dump sites are cluttered, shallow, turbid and full of snagging hazards, exactly the conditions where robots founder and a human hand reading a fuse by feel still wins. Munitions from two world wars are corroded into unfamiliar shapes, sometimes fused with sensitive mechanisms that tolerate no bumping; judging whether an item can be moved is expert intuition backed by ordnance knowledge. Legal accountability also lands on a named supervisor. So the trajectory is fewer dives per project rather than no divers, and the ones remaining are the hardest and most dangerous, which is precisely why our score stays low.
Automatability: our editorial assessment of current and near-term AI capability
The substitution is already underway and will keep advancing through the 2030s: expect most deep-water and offshore-construction clearance to be fully remote, with dive teams reserved for shallow, cluttered and archaeologically sensitive sites. Humanitarian programmes clearing coastal waters in post-conflict countries will lag on cost. Total diver numbers shrink slowly while the remaining work concentrates the difficulty. Nobody in the trade should expect to be automated out this decade.
Partly, and quickly, but not entirely. Remote vehicles already handle search, inspection and much charge placement in open water, which is the safest possible outcome. What they handle poorly is silted, cluttered, shallow water where manipulators snag and cameras see nothing. Those are exactly the sites near ports and beaches that most need clearing.
Yes. Offshore wind construction, cable and pipeline routes, port expansion and post-conflict coastal clearance all generate steady unexploded-ordnance work, and the seabed holds a century of accumulated munitions. Employment is contract-based and follows infrastructure spending. The composition of teams is shifting toward remote-system operators, so the growth is not evenly distributed among divers.
Substantially. Better sonar classification means fewer dives on scrap metal, autonomous survey covers ground no diver should cover, and remote intervention removes people from the blast radius on the majority of items. The paradox is that the dives that remain are the difficult ones, so individual dive risk stays high even as total exposure falls.
Cross-train into the systems replacing your bottom time: ROV piloting, sonar data interpretation, and survey planning. Add project supervision and the safety documentation that clients demand. Divers who can plan a remote-first clearance campaign and then personally handle the items the machines cannot touch will be the last people in the industry anyone lets go.