How AI is Unearthing Hidden Scientific Knowledge About Life on Earth

The Scale of Unknown Biodiversity

• 🌍 Scientists estimate that 80 percent of life on Earth remains unknown to humanity, with only two million out of an estimated ten million species observed.
• ⚠️ This lack of knowledge is critical as species are under threat from habitat loss, rising temperatures, and natural disasters, leading to extinction rates 100 to 1,000 times higher than expected.
• 🚨 Traditional data collection methods are too slow to keep pace with the current biodiversity crisis, making it difficult to understand species' needs and risks.

Leveraging Existing Ecological Data

• 💡 We are sitting on vast databases of ecological knowledge, such as iNaturalist with 300 million images, Xeno-canto with bioacoustic recordings, and camera-trap data.
• 🔍 Each image or recording contains a wealth of information beyond just species identification, including individual identification, social interactions, vegetation, and food sources.
• ⏳ Manually analyzing this data is infeasible; for example, it would take 40 years of full-time work to review all iNaturalist images.

AI-Powered Discovery with Inquire

• 🧠 AI researcher Sara Beery developed a system called Inquire to help ecologists discover knowledge hidden in data without needing to collect examples or code.
• 💬 Inquire allows scientists to directly ask questions to databases, comparing scientific language to millions of images in seconds to find relevant data.
• ✅ This system enables rapid identification of human-verified examples, significantly reducing the time and computational power needed for scientific discovery.

Accelerating Conservation and Research

• 📊 A study using Inquire found thousands of examples of bird diets in just three hours, a task that previously took teams 1,560 hours of manual curation.
• 🌲 The system has been used to explore diverse questions, such as forest regeneration after fires, species mortality differences, and changes in flowering events due to climate change.
• 🚀 Inquire can be extended to other data types like bioacoustic recordings and satellite imagery, paving the way for discovering hidden connections between different ecological data sources.

The Future of Conservation

• 🛠️ While AI tools alone cannot solve the biodiversity crisis, they maximize the value of existing data, enabling strategic resource allocation for new data collection.
• 🤝 By building complete pictures of life on Earth, AI tools empower scientists and the public to drive conservation actions and protect ecosystems more effectively.
• ✨ The future of conservation lies in both existing and future ecological databases, where AI can help us understand and protect the planet's biodiversity.