How Meta’s Llama AI model is being deployed to support sustainable development goals

A global intergovernmental organization is leveraging Meta’s Llama AI model to create an innovative chatbot designed to enhance public engagement with their Sustainable Development Goals (SDGs), marking their first major venture into public-facing generative AI.

Project Overview: Accenture has developed a proof-of-concept chatbot built on Llama 3.1 to help stakeholders better understand and engage with the organization’s extensive SDG-related content.

• The chatbot aims to serve diverse audiences, from policy experts to the general public, by making complex information more accessible
• The system can efficiently summarize lengthy reports and policy documents into digestible formats
• The implementation will enable stakeholders including business leaders, policymakers, educators, and citizens to track SDG progress and find ways to participate

Technical Architecture: The solution combines Llama 3.1 with AWS infrastructure and specialized tools to ensure robust performance and scalability.

• The system utilizes retrieval-augmented generation to ground responses in the organization’s content
• Llama Guard has been integrated to manage out-of-bounds queries and ensure appropriate responses
• The architecture was modified to handle the large 405B parameter model while maintaining cost efficiency

Key Features and Capabilities: The chatbot serves as a comprehensive knowledge interface for the organization’s sustainable development initiatives.

• Provides summaries and key points from extensive publications in user-friendly language
• Surfaces relevant report links within responses to guide users to additional resources
• Directs users to the organization’s website for deeper engagement with content

Future Development Plans: The project team has outlined several enhancements to expand the chatbot’s functionality and reach.

• Planning to evaluate smaller Llama models for comparable performance at reduced infrastructure costs
• Will expand language support to include French, Spanish, Chinese, Arabic, and Russian
• Aims to incorporate event information and develop multimodal capabilities for handling videos, graphics, and structured data

Strategic Implications: This implementation represents a significant step toward democratizing access to sustainable development information and fostering global collaboration through technology.

• The use of open-source LLM technology aligns with goals of promoting equitable access to innovation
• The solution helps bridge the gap between complex policy documents and public understanding
• Supports the organization’s mission to achieve its 2030 timeline through enhanced stakeholder engagement

Looking Ahead: The successful deployment of this chatbot could serve as a model for other international organizations seeking to leverage AI for public engagement, while its planned expansions into multilingual and multimodal capabilities suggest an evolving role in global development communication.