> ## Content Index
> Fetch the complete content index at: https://globalfeed.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# UN turns to Google to make its global statistics readable by AI systems
- URL: https://globalfeed.ai/en/un-turns-to-google-to-make-its-global-statistics-readable-by-ai-systems/
- Published: 2026-09-18T17:06:38.000Z
- Updated: 2026-09-18T17:06:38.000Z
- Description: The UN System Data Commons, built on Google's open-source platform, lets users query statistics from nearly 20 UN agencies in natural language and supports MCP.
- Author: GlobalFeed Editor
- Tags: Google, UN, UNICEF, x-Google, x-UN, x-UNICEF, dil-en, denetim-gerekli, video

The United Nations has launched a platform that pulls statistics from across its agencies into a form AI systems can query directly. Called the UN System Data Commons, it runs on Google's open-source Data Commons platform and accepts natural-language questions. It replaces the UNData portal, where users had to browse a more traditional database interface. The platform also supports the Model Context Protocol (MCP), which lets AI systems connect straight to external data sources.

Twenty-six UN entities have committed to the project, with data from nearly 20 available at launch. The UN wants 80% of the system's statistical datasets on the platform by 2027\. Google.org contributed $2 million in capacity-building funding plus technical support for the core infrastructure. Prem Ramaswamy, who leads Google's Data Commons team, said the system sits on a UN-governed instance and is meant to be maintained, operated and scaled by the UN itself over time.

## Accuracy of 21.2% in a UNICEF test

João Pedro Azevedo, UNICEF's chief statistician, said a benchmark of six large language models, covering more than 133,000 responses to questions about global development indicators, produced an average accuracy score of 21.2%. The test included OpenAI's GPT-4o and GPT-4o-mini, Anthropic's Claude Sonnet 4.5 and Haiku 4.5, and Google's Gemini 2.5 Flash and Gemini 2.0 Flash.

About three in five responses returned no usable number at all, often because the models hedged. When the same questions were run again on the same model versions roughly two days later, models that gave a number both times repeated the identical figure only about half of the time. The study is a UNICEF working paper that has not been peer-reviewed; the agency plans to publish its methodology, code and data with the paper.

UNICEF also reported a sharp rise this year in traffic from generative AI assistants to its data website, which draws more than six million visits a month. Between January 1 and September 14, visits from links in ChatGPT answers rose 67% year over year, accounting for 6.4% of all sessions; UNICEF estimates AI assistants overall now bring about one in every 10 visits.

The platform records the origin of each statistic, so a figure retrieved by an AI system can be traced back to the original UN source. Ramaswamy said models can misinterpret nuance and a person should review outputs before they are cited or published. Full report at [TechCrunch](https://techcrunch.com/2026/09/17/un-turns-to-google-to-make-its-global-data-ready-for-ai-agents/?ref=globalfeed.ai).

Second source: [Unite.AI](https://www.unite.ai/un-system-data-commons-launches-as-ai-ready-global-statistics-platform/?ref=globalfeed.ai).