■ SAFE RISK ■ Public Service & Government
Not replaced, though genuinely assisted. Machines clear vegetation and flag metal; a person still kneels in the dirt with a trowel to find out whether the signal is a mine or a bottle cap.
“Clearing landmines in Cambodia by hand. This is courage, not computation.”
Our AI replacement risk score — how we score jobs
Humanitarian demining is slow, methodical and physical. Deminers work a marked lane in body armour, sweeping a detector, then excavating each signal from the side at a shallow angle with a trowel and prodder because top-down pressure sets off pressure plates. Most signals are scrap. Deteriorated ordnance in tropical soil is unpredictable, unexploded submunitions behave differently from mines, and terrain — rice paddy, dense jungle, riverbank — dictates technique. Around the digging sits the rest of the job: community liaison to find out where devices actually are, risk education in villages, survey and mapping, medical standby, and the documentation that lets land be formally released.
Technology has arrived meaningfully. Drone surveys with thermal and multispectral imaging, and increasingly with ground-penetrating radar, help prioritise suspected areas. Machine analysis of imagery detects surface disturbance and craters. Armoured flails and tillers clear vegetation and sometimes detonate devices. Detection-dog and rat programmes are being augmented by sensor research, and data systems have transformed survey and land-release decisions.
What holds is verification. Nobody hands land back to farmers on a machine's say-so — quality assurance requires human excavation, and remote areas defeat heavy equipment. Community trust, which produces the information about where mines were laid, is relational. Our score of 6 reflects a field that is speeding up with technology while remaining staffed by people who accept personal risk on purpose. The people dimension matters more than outsiders expect. Villagers know where devices were laid, but they tell you only if they trust the organisation, and they will keep farming contaminated land regardless while waiting. Getting accurate information out of that situation is patient, local, human work.
Automatability: our editorial assessment of current and near-term AI capability
Resilient for the foreseeable future. Drone survey and automated imagery analysis are already compressing the search phase, and mechanical clearance handles vegetation and some detonation — expect steady gains through the 2030s that make each deminer far more productive. The global contamination backlog is enormous and growing with new conflicts, so productivity gains shorten timelines rather than eliminating jobs.
Machines already help. Armoured flails and tillers process vegetation and ground, drones survey suspected areas, and imagery analysis narrows where to look. But mechanical clearance leaves devices behind and damages some without detonating them, so land release still requires human verification with detector and trowel. In steep, forested, flooded or remote terrain, heavy equipment cannot operate at all.
Mainly in survey and prioritisation. Drone-captured thermal, multispectral and radar data is processed to flag ground disturbance, craters and probable contamination, which focuses expensive clearance on the right ground. Data systems track evidence for land release decisions. Research continues on improving detector signal discrimination to reduce the enormous share of excavations that turn out to be scrap metal.
No — contamination is growing. Legacy mines from decades-old conflicts remain across dozens of countries while recent wars are creating vast new contaminated areas. Funding fluctuates and is the real constraint on employment, but the technical need will outlast anyone's career. Productivity technology is welcomed by the sector because the backlog is measured in decades, not because it threatens staffing.
Core clearance competence and IMAS-standard accreditation remain the base. On top of that, the differentiating skills are technical survey — drones, GIS, imagery interpretation — plus community liaison, since local knowledge about where devices were laid is the single highest-value input. Team leadership and quality assurance roles are expanding as clearance becomes more equipment-intensive.