AI Below Sea Level

ChatGPT, Microsoft Copilot, and Gemini, once obscure technological terms, are now everyday essentials to millions of people. Artificial Intelligence now powers a multitude of tasks, from drafting emails to finding the best food spots or figuring out directions — it is a tool that can help us all. Beyond just productivity and convenience, AI is now increasingly venturing into complex scientific and environmental fields. Few could have predicted that AI could be used to help protect some of the planet's most vulnerable underwater ecosystems. As we race to save our seas from overfishing, coral bleaching, and pollution continue, AI is emerging as a powerful tool to give our oceans a fighting chance.

Led by the University of Plymouth and the Plymouth Marine Laboratory (PML), the $2 million project named the “Deep Vision Project” uses AI to construct large-scale, detailed databases that identify the most vulnerable underwater habitats across the Atlantic Ocean. This noninvasive AI uses advanced seabed mapping to scan large areas of the ocean floor to discover the ocean species that occupy these microenvironments. Through analysing thousands of images, the AI model learns to identify the presence of Vulnerable Marine Ecosystems (VMEs), which are fragile and slow-recovering ocean species that require urgent protection from further ecological damage. These include deep-sea cold-water coral reefs, sponge fields, and seamount ecosystems found in underwater canyons or trenches that the AI can detect despite the difficult terrain. The AI then uses these images to identify the specific temperatures these VMEs need to survive, enabling scientists to understand how to sustain these specific conditions for their long-term resilience and reproduction. Currently, the project is in Phase II, with fifteen micro-projects receiving additional grants to test the AI across various parts of the Atlantic. The ultimate goal is to upscale the AI to map global ocean patterns and safeguard all marine ecosystems for future generations.

This precise ocean data uncovered in the Deep Vision Project has another important purpose. This AI provides scientists and policymakers with tangible evidence to lobby for laws protecting these vulnerable ocean areas. The project team aims to strengthen advocacy and support for environmental management policies like the Biodiversity Beyond National Jurisdiction (BBNJ) treaty, which protects biodiversity in areas beyond official national jurisdiction as part of the United Nations Convention on the Law of the Seas. Further, the Deep Vision team also aims to showcase its AI research to investors, sparking broader interest in ocean protection among its more tech-focused funders. Supported by the Bezos Earth Fund — an organisation funding global climate solutions — the Deep Vision Project is backed by powerful organisations, demonstrating its powerful capacity to effect progressive reforms in ocean governance.

‘‘As the deep sea becomes more accessible for exploitation, there is an urgent need to map its ecosystems to enable the conservation of biodiversity in this, the last wild frontier on Earth.’’ This statement by Kerry Howell, University of Plymouth professor and leader of the Deep Vision project, reflects the critical importance of innovative approaches to protect our ocean ecosystems amid growing environmental threats. It is fascinating to explore how we can use AI to help combat this threat. Beyond just helping us make lists or check for grammar, AI can empower scientists to tackle some of their greatest environmental challenges.

Illustration by Iona Talbot Rice