■ SAFE RISK ■ Science & Research
The surveying is going robotic; the planting is not. Attaching a fragile coral fragment to irregular substrate in surge remains a job for hands, and funding for reef work is growing rather than shrinking.
“Coral planting is delicate underwater work. Robots are still too clumsy.”
Our AI replacement risk score — how we score jobs
The work is field biology performed on scuba. Maintaining nursery structures — tree and table arrays where coral fragments grow out — cleaning them of algae and predators, fragmenting healthy colonies, then outplanting onto degraded reef using epoxy, cement or nails, and returning to monitor survivorship. Add transect surveys, photo quadrats, water quality sampling, crown-of-thorns or lionfish removal, and the substantial safety and logistics overhead of repetitive dive operations from small boats.
Autonomy has taken the data side quickly. Photogrammetry rigs and AUVs map reefs in three dimensions, and vision models classify coral cover, bleaching state and species composition from imagery far faster than a diver with a slate. Some larval-dispersal and substrate-seeding robots exist and are genuinely useful for scale-critical restoration. Environmental DNA sampling reduces the number of survey dives needed. Data pipelines, reporting to funders and permit compliance are all being handled by software that did not exist a decade ago.
Where machines stop is the manipulation. Reef substrate is irregular, biofouled and surging; a coral fragment is brittle and needs firm contact without tissue damage; the diver decides in seconds whether this patch of dead plate is stable enough to hold. Underwater manipulators capable of that dexterity are expensive, slow, and mostly built for oil and gas. Meanwhile demand is rising with reef decline, coastal protection funding and carbon-adjacent finance. Our risk score of 21 reflects survey work moving to autonomy while the restoration itself stays human, with the field's real constraints being funding cycles and whether reefs survive long enough to restore.
Automatability: our editorial assessment of current and near-term AI capability
Survey work is already shifting to AUVs and vision models, and that will be near-complete by 2030. Outplanting stays manual well past 2040 barring a breakthrough in cheap underwater manipulation. The sector's growth depends on climate finance and coastal protection budgets, both trending upward. Expect divers to spend proportionally less time collecting data and more time on restoration and larval-based techniques.
Yes, mainly for mapping and for larval dispersal or substrate seeding at scale. Autonomous vehicles survey reefs far faster than divers and vision models classify the imagery well. Attaching individual fragments to irregular substrate in surge is the part that has stayed human, because the required dexterity is expensive and slow to engineer underwater.
Growing, though funding-dependent and often seasonal. Reef decline, coastal protection value and climate finance are all pushing money into restoration, and organisations need people who can dive competently and handle data. Job security tracks grant cycles more than technology, so versatility across survey, nursery and outreach work matters.
More restoration and less counting. If image classification handles cover estimates and bleaching assessment, diver time reallocates to outplanting, nursery work, larval deployment and intervention experiments. That is arguably a better job — the tedious transect data collection is exactly the part divers complain about, and it is the part automating.
Scientific or commercial diving certification beyond recreational level, a marine biology or ecology background for most research roles, and practical boat and equipment competence. Increasingly employers want image analysis or GIS skills too. Volunteering and field seasons are the standard entry route, which makes the early years financially thin.