2024-11-03-UNDP-利用人工智能加强昆明-蒙特利尔全球生物多样性框架的早期行动（英）_36页_3mb

NBSAP Target Similarity Assessments: Summary

Context

Countries must align their National Biodiversity Strategies and Action Plans (NBSAPs) with the Kunming-Montreal Global Biodiversity Framework (GBF) to achieve biodiversity targets. However, capacity gaps and delays hinder progress.

Objective

UNDP's Early Action Support (EAS) project uses human-centered AI (GPT-3.5) to evaluate the similarity between national NBTs and GBF targets, accelerating policy alignment and stakeholder engagement.

Methodology

1. Data Collection: Publicly available NBTs from the CBD database.
2. AI Model: GPT-3.5 analyzes text similarity between NBTs and GBF goals/targets.
3. Validation: Results are reviewed by national experts to ensure accuracy and context relevance.

Key Findings

Took place of forming groups and subgroups. Summarized the situation where boxes are used or not, and what is the next step. Drew lines markings for the ones that have similarities or are different.

Global Trends

AI was used to analyze 3,065 pre-COP15 NBTs, revealing that topics like wild species management were well-represented, while gender equality and urban planning were underrepresented.

Benefits

1. Accelerated target alignment processes in 54 countries.
2. Democratized access to policy analytics.

Challenges and Mitigation

1. Bias risks: Human oversight and expert review to contextualize AI outputs.
2. Environmental impact: Minimal data input to reduce computing needs.
3. Limited scope: Future models to include implementation gaps and harmonization with human rights frameworks.

Lessons Learned

1. AI is a powerful tool but must be combined with traditional expert input.
2. Focus on specific actions beyond text similarity to avoid redundancy.

Future Directions

Explore AI for cross-sector policy synergies and financial planning in biodiversity conservation.