UN taps Google to make its global data AI-ready

The United Nations said Thursday it is working with Google to make its vast collection of global statistics easier for AI systems to access and use. The new UN System Data Commons is built on Google’s open-source Data Commons platform and lets people query UN statistics using natural language, replacing the existing UNData portal, which relied on a more traditional browse-and-search database interface. The new platform also supports the Model Context Protocol (MCP), a standard that lets AI systems connect directly to external data sources.

The launch comes alongside evidence of how unreliable AI systems still are with this kind of data. A UNICEF benchmark of six large language models across more than 133,000 responses to questions about global development indicators produced an average accuracy score of just 21.2%, João Pedro Azevedo, the agency’s chief statistician, told reporters. The test covered 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 didn’t provide a usable number at all, often because the models hedged, Azevedo said — and when the same questions were rerun on the same models about two days later, models that gave a number both times returned the identical figure only about half the time. The study is an unpublished UNICEF working paper that hasn’t been peer-reviewed.

UNICEF has also seen a sharp rise in traffic from generative AI assistants to its own data website, which gets more than 6 million visits a month: referrals from users clicking ChatGPT links rose 67% year-over-year between January 1 and September 14, Azevedo said. The UN said 26 of its entities have committed to the new Data Commons, with data from nearly 20 available at launch, and it aims to bring 80% of the UN system’s statistical datasets onto the platform by 2027. Google.org provided $2 million in funding and technical support to build the platform’s core infrastructure, which is hosted on a UN-governed instance that Google intends the UN to eventually run independently.

Beyond individual statistics, Google showed how an AI system connected via MCP could pull together several indicators at once to generate dashboards, charts, and written analysis — in one demonstration, an AI system used UN data on HIV infections, AIDS mortality, and life expectancy to produce an infographic on the impact of the US President’s Emergency Plan for AIDS Relief in Africa. Even so, Google’s Prem Ramaswami, who leads the Data Commons team, cautioned that authoritative source data doesn’t guarantee authoritative conclusions: “Because models can misinterpret nuance, a human should always review the outputs before citing or publishing them,” he said.

Why it matters

With AI assistants already driving a meaningful and fast-growing share of traffic to UNICEF’s own data properties, a 21.2%-accuracy baseline is a reminder that “AI-ready” data infrastructure solves only half the reliability problem — the other half still runs through the model doing the interpreting.