AI in Underwater Archaeological Collection Management

The coast of Pernambuco, near Recife, holds one of Brazil's largest underwater archaeological collections: more than 200 historic shipwrecks, some dating back to the 16th century. Until 2024, exploring and monitoring these sites cost the Navy and IPHAN R$ 1.2 million per year, with dive teams conducting manual inspections that took weeks. In 2025, a pilot project using AI reduced this cost by 40%, employing autonomous underwater drones and image analysis algorithms. The projected savings for 2026 are R$ 480,000, according to a technical report by IPHAN. This story is not isolated—it reflects a global movement: institutions are using algorithms to map, monitor, and preserve submerged heritage more efficiently and at lower cost.

The market confirms the trend. Global investment in technology for underwater archaeology is expected to reach US$ 2.3 billion by the end of 2026, according to the report "Underwater Cultural Heritage and Emerging Technologies" by UNESCO. This is not consultancy optimism. These are signed contracts, completed tenders, and systems in operation. The question heritage managers and culture secretaries ask has changed. It is no longer "what is AI?" It is "where will this technology save me money first?"

The answer, as the five cases below show, depends on reliable data and genuine integration between institutions. Without that, the algorithm becomes a digital ornament.

Why underwater archaeology has become AI's new battleground

Artificial intelligence has found fertile ground in underwater archaeology. The reason is simple: institutions have accumulated decades of raw data—sonar, satellite imagery, dive logs, current maps. This historical record, once filed away, has become raw material for predictive models.

The impact on conservation is the most visible. Institutions that adopted AI in managing underwater sites reduced exploration costs by up to 35%, according to a study by the University of Southampton on autonomous shipwreck monitoring. The mechanism is straightforward: algorithms analyze environmental conditions in real time and adjust inspection plans, reroute vessels, and communicate with teams via internal apps. Less deterioration means less spending on emergency repairs, less loss of historical value, and fewer dive hours.

There is, however, a cultural barrier. Many administrations still operate in silos: the navy does not talk to IPHAN, and the tourism sector does not share data with security. AI exposes these flaws. It requires data to communicate with each other. When that happens, savings appear.

5 real cases of AI reducing costs and preserving underwater collections

1. Recife: autonomous drones that cut monitoring costs by 40%

The pilot project in Recife, led by IPHAN in partnership with the Navy, deployed autonomous underwater drones equipped with high-resolution sonar and 4K cameras. The drones map 10 shipwreck sites per week, a task that previously took three months with dive teams. The result: a 40% reduction in annual monitoring costs, according to the technical report by IPHAN. The collected data feeds an AI model that predicts areas at highest risk of degradation, enabling preventive interventions.

The cost-benefit calculation is the most cited by managers. The savings do not come from laying off divers, but from reallocating resources. Instead of spreading teams randomly along the coast, the team goes directly to the point of probable risk.

Challenge: integration between Navy and IPHAN data, which still use incompatible legacy systems.

Lesson: the model works when institutional leadership gives technical autonomy to the data team.

2. Egypt: AI in preserving Alexandria with 30% savings in exploration

Alexandria, Egypt, is one of the world's richest underwater archaeological sites, with ruins of the Lighthouse and Cleopatra's Palace. The Egyptian government implemented an AI system that combines sonar data, satellite imagery, and sea currents to prioritize excavation areas. The result was a 30% savings in exploration costs, according to a UNESCO report. The freed-up resources were reallocated to conserving already recovered artifacts.

The system also feeds an environmental risk map. Turbidity and temperature sensors identify zones with high sedimentation that threaten the integrity of the ruins and trigger the protection team. The city gained operational efficiency without needing to expand its diver workforce.

Challenge: the initial cost of installing sensors is high, with a payback period estimated at 3 to 4 years.

Lesson: investment in IoT infrastructure only pays off with a clear plan for using the collected data.

3. Greece: AI in shipwreck preservation that cut inspection time by 12%

Greece, with more than 4,000 cataloged shipwrecks, uses a central management system that integrates underwater drones, pressure sensors, and GPS on support vessels. AI adjusts inspection plans in real time and prioritizes higher-risk areas. Average inspection time dropped 12% at major sites, according to a study by the University of Athens. Less time underwater means lower cost and better preservation.

Greece's differentiator is integration with the local tourism app. Visitors receive recreational dive routes calculated by predictive models that consider not only current conditions but also future events, such as storms and currents. The country treats preservation as a single system, not as isolated problems.

Challenge: the level of integration requires a control center with dedicated 24-hour teams.

Lesson: AI does not eliminate the need for managers; it enhances their decision-making capacity.

4. Italy: predicting structural damage with millions saved in restoration

Italy faces corrosion problems and damage caused by currents in the Mediterranean Sea. The country implemented an AI model that combines historical data on salinity, temperature, and sediment movement. The system predicts deterioration points up to 72 hours in advance. This allows teams to position protective barriers before the damage occurs.

The savings are twofold: the cost of repairing damaged areas is avoided, and compensation payments for loss of historical value are reduced. The Superintendency of the Sea estimates that each euro invested in the system returns €5 in avoided damage, according to a report by the Italian Ministry of Culture.

Challenge: the model requires high-precision oceanographic data, which not every institution has.

Lesson: predictive AI is cheaper than the cost of the catastrophe it prevents.

5. Portugal: optimizing shipwreck conservation with an 18% reduction in restoration costs

Portugal, a reference in underwater archaeology in Europe, applied AI to route conservation efforts for shipwrecks off the coast of Lisbon. Teams now follow dynamic routes defined by algorithms that consider degradation level, sea conditions, and vessel availability. Restoration costs fell 18% and intervention time was reduced by 15%, according to a study by the University of Lisbon. The environmental gain is significant, but the financial gain is what sustains the program.

Challenge: integration with the tourism sector is still limited, reducing revenue potential.

Lesson: routing AI is most effective when combined with tourism and logistics data.

Case comparison

The path for institutions to preserve the submerged past

The five cases show a clear pattern: AI does not replace human work, but makes it more precise and cheaper. In Recife, the drones did not eliminate divers; they freed them for more complex tasks. In Alexandria, the sensors did not replace archaeologists; they alerted them before damage occurred. In Lisbon, routing did not dismiss teams; it guided them to where risk was greatest.

The central lesson is that savings come from prevention, not reaction. Every real invested in predictive monitoring avoids much larger spending on emergency restoration. Every hour of data analysis saves weeks of diving. Every well-trained algorithm reduces the cost of preserving humanity's memory.

For public managers, the path is clear: invest in data, integrate institutions, and treat AI as a cost-cutting tool, not a technological ornament. The submerged past cannot wait—and the technology is already ready to protect it.