The United Nations announced on Thursday that it is collaborating with Google to make its international statistical databases accessible to artificial intelligence systems through a new platform called the UN System Data Commons. Built on Google’s open-source Data Commons infrastructure, the initiative allows users to query statistics across UN agencies using natural-language search. Google.org supplied $2 million in capacity-building funding along with technical support to construct the platform’s core infrastructure.
The move follows findings from a UNICEF study that revealed major large language models struggle to accurately retrieve global development indicators. In a benchmark test evaluating six LLMs across more than 133,000 responses, the models achieved an average accuracy score of 21.2%, according to João Pedro Azevedo, chief statistician at UNICEF.
The benchmark evaluated 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. Roughly three out of five responses failed to return a usable figure, frequently due to model hedging. Furthermore, when identical questions were resubmitted to the same model versions approximately two days later, models that supplied numerical outputs on both runs provided matching numbers only about 50% of the time. The evaluation is documented in a UNICEF working paper slated for journal submission, with methodology, code, and underlying data set to be released publicly.
The technical overhaul coincides with increased traffic from AI tools to official UN resources. UNICEF reported that visits to its data site—which draws over 6 million monthly visits—originating from ChatGPT answer links grew 67% year-over-year between January 1 and September 14, 2026. Referral links from ChatGPT accounted for 6.4% of all site sessions during that period, with overall generative AI assistants estimated to represent roughly 10% of total site traffic.
According to the UN, 26 of its entities have committed to participating in the UN System Data Commons, with data from nearly 20 available at launch. Shantanu Mukherjee, acting director of the UN Statistics Division, stated that the initiative marks the first time data has been connected at this scale across the UN system while simultaneously structuring the records for AI retrieval.
Sources: [TechCrunch](https://techcrunch.com/2026/09/17/un-turns-to-google-to-make-its-global-data-ready-for-ai-agents/)
Source: TechCrunch
Corrections
If you believe this article contains an error, contact Herald Express with the source URL and supporting evidence.
Story synopsis gathered from: TechCrunch — source
