UN turns to Google to make its global data ready for AI agents
News Source
β’Thu, 17 Sep 2026 20:00:00 +0000
π° What Happened
The United Nations said on Thursday it is working with Google to make its huge collection of global statistics easier for AI systems to find and use. The new system is called the UN System Data Commons.
It is built on Googleβs open-source Data Commons platform. People can ask questions in plain language instead of clicking through a database. The system replaces the older UNData portal, and it also supports a standard called the Model Context Protocol, which lets AI systems link straight to outside data.
π The Backstory
UN agencies publish masses of data on health, poverty, food and trade. The old site made users browse and search by hand. That is slow for people and hard for AI tools to read.
A UNICEF test showed how bad the problem is. Six popular AI models answered more than 133,000 questions about world development numbers. Their average accuracy was only about 21 percent, and about three in five answers gave no usable number at all. When the same questions were run again two days later, models that gave a number both times matched only about half the time.
π― Why It Matters
If AI tools give wrong numbers about poverty or health, then plans, news reports and school projects can all end up wrong too.
The United Nations said on Thursday it is working with Google to make its huge collection of global statistics easier for AI systems to find and use. The new system is called the UN System Data Commons.
It is built on Googleβs open-source Data Commons platform. People can ask questions in plain language instead of clicking through a database. The system replaces the older UNData portal, and it also supports a standard called the Model Context Protocol, which lets AI systems link straight to outside data.
UN agencies publish masses of data on health, poverty, food and trade. The old site made users browse and search by hand. That is slow for people and hard for AI tools to read.
A UNICEF test showed how bad the problem is. Six popular AI models answered more than 133,000 questions about world development numbers. Their average accuracy was only about 21 percent, and about three in five answers gave no usable number at all. When the same questions were run again two days later, models that gave a number both times matched only about half the time.
If AI tools give wrong numbers about poverty or health, then plans, news reports and school projects can all end up wrong too.