A critical component of the launch is the platform’s support for the Model Context Protocol (MCP), a technical standard that enables AI agents to connect directly to external data sources to retrieve verified information.
This transition addresses a significant reliability gap in current AI infrastructure, as UNICEF research found that leading large language models—including versions of GPT, Claude, and Gemini—averaged only a 21.2% accuracy rate when answering questions about global development indicators.
By integrating UN data into a standardized framework, the partnership aims to prevent "hallucinations," or fabricated responses, by ensuring AI tools can cite authoritative sources.
This is increasingly vital as UNICEF reports that AI assistants now drive approximately one in ten visits to its data websites, representing a 67% increase in referrals from tools like ChatGPT over the past year.
The UN intends to have 80% of its statistical datasets on the platform by 2027, with Google providing $2 million in funding and technical training to ensure the UN can eventually operate the system independently.
While the technology allows AI to automatically generate complex dashboards and infographics from raw data, Google representatives cautioned that human oversight remains necessary because models may still misinterpret nuances in the information.
To assist with transparency, the platform includes tracking mechanisms so every statistic an AI retrieves can be traced back to its original agency source.