AI solution for IUCN’s resolutions & recommendations management

IUCN supports scientific research, manages field projects all over the world and brings governments, non-government organizations, United Nations agencies, companies and local communities together to conserve nature, promote sustainable development, and create a just world where nature is valued and protected.

S-PRO partnered with IUCN to implement an AI solution for managing Resolutions and Recommendations, enhancing access to environmental policy insights. Launched in Summer 2024, this solution leverages over 1,466 archived documents, including 695 active ones, to provide IUCN staff with an efficient, user-friendly interface for quickly accessing and summarizing key information.

Complex Document Landscape

One of the main challenges was managing IUCN’s extensive repository of over 1,466 Resolutions and Recommendations, with 695 still active. The AI needed to recognize relationships between active and archived documents, understand how policies evolved through amendments, and accurately track which directives were current. This required the AI to not only process documents individually but also synthesize information cohesively, referencing the latest and most relevant documents to ensure accurate summaries.

Intelligent Summarization and Contextual Awareness

Given the layered and interconnected nature of the Resolutions and Recommendations, training the AI to recognize patterns, thematic clusters, and historical context required advanced natural language processing capabilities. The AI needed to accurately identify relevant themes and cite specific documents to support summary outputs. This required developing an intelligent model capable of analyzing document clusters without losing context or generating summaries that inadvertently misrepresent the intentions of IUCN’s policy evolution.

Balancing Advanced Technology with a User-Friendly Experience

Another key challenge was creating a powerful yet intuitive interface that would allow IUCN staff to fully leverage the AI’s capabilities without requiring extensive training. While the back-end AI involved complex functionalities, such as filtering by topic or geography and mitigating potential AI-generated “hallucinations,” we aimed for a streamlined, chat-like interface. This interface had to enable staff to query, retrieve, and synthesize policy information with ease, making sophisticated AI technology feel accessible and natural to use.

The AI solution developed for IUCN focuses on summarizing and synthesizing key policy documents, allowing users to query specific themes across multiple resolutions and recommendations. This system groups policies into thematic clusters – such as oceans, climate change, and governance – and offers an interactive chat interface for intuitive access.

The technology stack was designed to align with IUCN’s infrastructure requirements and data security standards:

Microsoft Azure

Provides a secure hosting environment, ensuring IUCN’s data remains protected and accessible at all times.

Azure OpenAI and Cognitive Search

Streamlines document indexing, embedding, and retrieval, making it easier for staff to access and work with resolutions and recommendations.

Hugging Face Chat UI

Delivers a user-friendly interface to facilitate human-oriented interactions with the AI model.

Machine Learning Models

Various ML models enable accurate document clustering, classification, and policy extraction, improving the relevance and usability of the content.

The AI solution accurately tracks policy changes, ensuring teams always had the most current insights and could assess the impact of new policies. The user-friendly chat interface requires minimal training, promoting easy adoption and increasing productivity.

Enhanced data availability and better context provides a solid foundation for future policy analysis and development. This solution met IUCN’s initial goals by improving data retrieval, supporting clear policy insights, and creating a scalable framework for future AI initiatives.

The AI capabilities extend to natural language processing for contextual understanding and summarization, providing detailed citations for all summaries generated. Key features include:

Document Clustering

Organized thematically, enabling quick access to policies relevant to specific environmental issues.

AI-Powered Interface

Users interact with a chatbot-like interface to query topics, receive summaries, and access detailed insights.

Contextual Policy Summarization

AI extracts and summarizes key points with precise references, supporting detailed analysis and improved decision-making.

Future-Ready

The solution’s structure is scalable for future enhancements, including broader document databases and more sophisticated analytical capabilities.

The AI solution for IUCN improved how documents are accessed and analyzed, helping their team quickly find relevant information and stay updated on policy changes. Partnering with IUCN was a mutually rewarding experience, blending their expertise in environmental policy with our technological innovation. Together, we created a solution that supports meaningful action on global environmental challenges, and we’re proud to have contributed to this important mission.