By Idowu Bankole

In a world increasingly focused on sustainability, one Nigerian researcher is researching a new suite of tools to protect the planet using Artificial Intelligence (AI). Dr. Kehinde Owoeye, a Nigerian researcher based in England, is at the forefront of applying AI to solve environmental problems, from conservation to understanding complex ecosystems. Vanguard had a chat with him about his work in this area and the implications for Nigeria.

Fighting for the Birds: A Mission to Save Endangered Migratory Species

Dr. Owoeye’s passion for conservation is rooted in a startling reality. “Every year, an estimated 100 million to 1 billion birds die from fatal collisions with man-made structures like buildings, high-tension lines, and wind turbines,” he explained. “This raises a huge concern for conservationists”. He believes that by using AI to model migration patterns, we can reduce this tragic loss of life and help achieve key sustainable development goals.

In a recent work yet to be published, he developed a model that forecasts the movement of birds, such as the Turkey Vulture. His recent research combines traditional geospatial environmental data with a novel source: Large Language Models (LLMs).

“The data we collect from satellite tags is useful, but we found that LLMs like GPT-4o could provide even more expressive and useful information about a location’s features, like economic activity,” he said. He discovered that this combination of data gives artificial intelligence models a significant edge, outperforming models that rely on a single data source.

Unlocking Nature’s Secrets: The Food Web

Beyond birds, Dr. Owoeye’s research also delves into the intricate relationships that exist within ecosystems. He views the food web as a critical tool for understanding how different parts of nature are connected.

“The relationships in an ecosystem are all intertwined,” he noted. “For instance, climate change can lead to biodiversity loss, which in turn affects food security and human health”. He stressed that to truly address sustainability, we can’t look at problems in isolation.

To study these connections, he built a model that can learn not only if a relationship exists between species in a food web, but also the nature of this relationship, such as if they are predators or parasites. He also tackled a major challenge: a lack of high-quality data. By developing a technique, he calls “quasi-data augmentation,” he was able to improve the model’s performance, even with limited datasets.

This research has practical implications for ecosystem restoration. His work helped identify “keystone species”—organisms that are crucial for the stability of an ecosystem.

“This is a major development,” he told us. “It shows that future conservation efforts should be focused on protecting these key species to ensure the overall health of the ecosystem”.

Implications for Nigeria

Nigeria’s ecosystems, particularly the Niger Delta’s coastal areas and inland forests, have suffered extensive damage. The AI tool can be a game-changer for restoration projects by providing data-driven insights for instance, in prioritising restoration efforts as well as well as aligning environmental protection strategies with broader Sustainable Development Goals (SDGs).

Also, protecting Nigeria’s wildlife, especially endangered species, becomes more effective with AI. The technology can process data from camera traps and acoustic sensors installed in national parks. It can identify individual animals, track population movements, and even detect specific sounds like gunshots or vehicle engines associated with poaching. This capability allows for immediate alerts to park rangers, indicating the precise location of a threat.

Challenges Ahead

While AI holds significant promise in this area, there are challenges. In his words,

“There’s the issue of scarcity and quality. For our food web models, for example, we faced a major obstacle: ‘the lack of high-quality data detailing the actual relationships between species.’ Field collection is time-consuming and expensive, so to overcome this, we developed a novel technique for data augmentation This allows us to essentially enhance our limited datasets, significantly improving the model’s performance without requiring years of new field research.”

These datasets can also be very expensive to collect. According to him “this high cost of data acquisition acts as a significant barrier to entry, particularly for research institutions or start-ups in developing nations, further exacerbating the data disparity between the global North and South.”

For a nation like Nigeria, there is also the problem of lack of data centres and compute infrastructures. According to him, “Modern AI algorithms are compute-intensive. Training and running inference on these massive algorithms require immense processing power which Nigeria lacks at the moment unlike the developed countries.”

Looking Ahead: AI for a Better World

For Dr. Owoeye, this is just the beginning. He believes that AI will play an increasingly important role in our collective effort to build a sustainable future.

“The AI models we’ve built are just a small part of a much bigger puzzle,” he concluded. “My hope is to continue to use intelligent technologies to help build a sustainable future and provide a more holistic understanding of how our actions impact the environment”.