The Transformative Potential of AI for Wildlife Conservation
Image by Author using Microsoft Image Creator
Introduction
This article explores artificial intelligence’s (AI’s) remarkable potential to improve wildlife conservation. The image above mirrors my hopefulness about that potential. But, as addressed in this article, concerns exist about AI’s use in wildlife conservation. The question in this context, as in others, is not so much whether AI will behave itself but rather whether we’ll behave ourselves using it.
Limitations of Traditional Pre-AI Conservation
Despite its importance, wildlife conservation has faced numerous challenges throughout history. Traditional pre-AI conservation has been limited by several factors:
Time-consuming data collection: Manual tracking and monitoring of wildlife populations are labor-intensive, time-consuming, and often overwhelming. Trying to personally count members of quickly passing flocks or herds is reminiscent of the iconic I Love Lucy candy factory conveyor belt episode.
Limited coverage: Many species are naturally elusive (how can you blame them?) or inhabit remote, inaccessible areas, making traditional monitoring methods ineffective or impractical.
Human error: Manual data collection and analysis are susceptible to human error and bias, leading to misguided conservation efforts.
Cost: Conservation projects are often expensive, requiring funding for equipment, personnel, and operational costs. Limited financial resources can constrain the scope and effectiveness of conservation efforts.
Real-time monitoring challenges: Traditional methods often must rely on lagging snapshots rather than real-time data to respond rapidly to threats.
Data analysis bottlenecks: Processing large amounts of collected data can be time-consuming and resource-intensive. Traditional methods — relative turtles — lack the speed and efficiency to process large datasets, leading to delays in decision-making.
Limited Resources for Anti-Poaching Efforts: Combating poaching has been a constant struggle, with conservation organizations often needing more resources to patrol vast protected areas effectively.
Examples of AI’s Role in Improving Wildlife Conservation
Here are just a few of the remarkable examples of AI being used to improve wildlife conservation:
Wildlife Populations and Distribution: Google’s Deep Mind collaborated with Tanzania's Serengeti National Park to develop an AI model that can analyze camera trap footage, automatically tracking wildlife populations, allowing researchers to gain valuable insights with vastly accelerated speed and efficiency. This collaboration helps conservationists in the Serengeti better understand animal populations and distribution, leading to more informed decisions for protecting this vital ecosystem.
Wildlife Identification: Wild Me’s “Flukebook” project is the functional equivalent of a digital fingerprint scanner for whales and dolphins. This innovative program utilizes AI to analyze vast collections of whale and dolphin photos. AI can automatically identify individual animals by focusing on the unique patterns on their tails (flukes). This streamlines a previously time-consuming process and allows more comprehensive tracking of these marine mammals. The improved data collection leads to a deeper understanding of their populations and migration patterns, and potential threats these species face.
Wildlife Migration: BirdNet, an app that identifies birds by their sounds, collects user recordings that AI analyzes. This data helps researchers, including a flock of citizen scientists, track bird migrations by pinpointing locations and species along their journeys, contributing to valuable scientific insights.
Poaching: TrailGuard AI uses artificial intelligence to combat illegal poaching by deploying hidden cameras equipped with AI to immediately detect humans and animals. This network of digital eyes, placed in strategic locations within national parks and wildlife reserves, uses AI to analyze the footage and send instant alerts to park rangers when poachers are detected. This enables rapid response to potential threats, significantly improving the ability to protect endangered species from illegal activities.
Biodiversity: Wildlife Insights, a Wikipedia of wildlife data, uses AI to analyze camera trap data, identify species, and track populations. This data helps prioritize conservation efforts for high-biodiversity areas and fosters collaboration among researchers, leading to better wildlife protection on a global scale.
Wildlife Trafficking: In collaboration with various conservation organizations, Microsoft’s AI for Earth program has developed SEEKER, an AI system to combat illegal wildlife trafficking. SEEKER uses computer vision and machine learning to rapidly scan through luggage X-ray images at airports and border crossings, identifying potential wildlife products or live animals.
Protecting and Managing Habitats: The Rainforest Connection (RFCx) has developed an AI-powered system called "Guardian" to monitor and safeguard rainforest habitats. Akin to a benevolent eavesdropper, this system uses repurposed smartphones placed in trees to continuously record ambient sounds. AI algorithms then instantly analyze these audio streams to detect signs of illegal logging, such as chainsaws or trucks, and to identify and track various animal species. This allows rapid response to threats and provides valuable biodiversity and ecosystem health data.
Monitor Wildlife Health: The World Wildlife Fund’s (WWF) "Eyes on Recovery" project utilizes AI-powered camera traps to monitor wildlife affected by the devastating Australian bushfires. The AI system automatically filters out irrelevant images, quickly identifying and categorizing wildlife, allowing researchers to track various species' health and recovery efficiently. This rapid data processing ensures timely interventions and supports the restoration of ecosystems impacted by the fires.
Human-Wildlife Conflict Mitigation: The Elephants and Bees Project utilizes elephants’ natural aversion to African honey bees to resolve human-wildlife conflict. The project, a scarecrow for elephants, deters elephants from raiding crops by strategically placing beehive fences around farmland borders. This not only protects farmers’ livelihoods but also minimizes harm to the elephants themselves. This innovative approach fosters peaceful coexistence between humans and these magnificent creatures.
Climate Change: One of the best examples of AI being used to combat climate change in ways beneficial to wildlife is the World Wildlife Fund-Netherlands’ collaboration with AI experts to develop a predictive model for deforestation. This AI tool, piloted in Borneo and Gabon, can forecast forest loss up to six months in advance with 80% accuracy. By predicting deforestation, the model helps local communities and conservationists take proactive measures to protect critical habitats for wildlife, thereby mitigating the adverse effects of climate change on biodiversity.
Decoding Animal Communications: See the June UUAM newsletter.
Concerns about AI-Powered Wildlife Conservation
While AI offers tremendous potential, some serious concerns exist, including, among others:
Harmful Wildlife Conservation Efforts: Some government wildlife conservation/management efforts are, in the opinions of many animal rights and welfare advocates, cruel, misguided, counterproductive or exploitative rather than helpful to wildlife. Sadly, using AI to enable those efforts will exacerbate the harm inflicted on wildlife.
Climate Impact: AI systems, particularly large language models and deep learning networks, require significant computational power, leading to substantial energy consumption and associated carbon emissions. Additionally, the cooling systems for data centers housing AI infrastructure use large amounts of water, potentially exacerbating water scarcity issues in some regions. These climate impacts cast a shadow on the sustainability of AI as a conservation tool and leave one worrying that Big Tech’s contributions to AI-powered conservation are greenwashing.
Poachers and hunters: AI technologies such as advanced cameras, drones, and predictive algorithms will be misused by an increasing number of poachers and hunters to track and locate wildlife more effectively, thereby further exposing the myth of “fair chase” hunting.
Techno-Solutionism Trap: Techno-solutionism — comparable to treating the symptoms of a disease but not its cause — is the quixotic belief that advanced technology, such as AI, is a silver bullet that can eradicate complex social, political, economic, and environmental problems without addressing underlying systemic issues.
Data Quality and Bias: AI algorithms are trained on data sets. If they are inaccurate, biased or unrepresentative — “garbage in”—the results will likely be invalid — “garbage out.”
Ethical Considerations: In echoes of Big Brother, using AI to monitor wildlife raises ethical questions about wildlife privacy and the potential for misuse of technology.
Accessibility and Cost: While AI offers advantages, the technology and expertise required can be expensive.
Technical Limitations: AI technology requires significant computational power and technical expertise. Developing and maintaining AI systems can be resource-intensive, posing a challenge for underfunded conservation projects
Implementation and Accessibility: A major hurdle is ensuring that AI technologies are accessible and usable by conservationists in developing regions. Training and support are needed to enable effective AI implementation.
Integration with Existing Methods: Integrating AI with traditional conservation methods requires careful planning and coordination. Ensuring that AI complements rather than replaces existing strategies is essential for holistic conservation efforts.
Interdisciplinary Orchestration: Effective orchestration among diverse fields like conservation organizations, park rangers, wildlife biologists and ecologists, computer scientists, researchers, academics, Big Tech, climate scientists, ethicists, legal experts, GIS experts, roboticists, veterinarians and creators of automated cameras, acoustic recorders, GPS collars, drones, satellites and other tools is critical to successfully using AI for wildlife conservation. AI, by itself, cannot do the job.
Conclusion
We should approach AI-powered wildlife conservation with both genuine hope and (dare I say) optimism, but also with wide-eyed caution. AI is not Noah’s Ark for the digital age, magically saving wildlife from the flood of human and other threats. But properly used, it’s a powerful tool with the potential to revolutionize how we protect wildlife. Our challenge is to ensure we use it properly.
This article is based on many scholarly papers, news articles, podcasts and other information, including the items linked in the article and those recommended below.
-Frank Brown, UU of Arlington, Virginia
Further Reading:
Artificial Intelligence Is Watching Wildlife
How artificial intelligence is changing wildlife research
How AI is helping scientist protect birds
From Radar to AI: The future of conservation
Rangers Use Artificial Intelligence to Fight Poachers
5 Ways Regular People Are Tracking Wildlife With Personal Tech
How artificial intelligence buys valuable time to protect wildlife