UN Collaborates with Google to Enhance Data Accessibility for AI Agents

The United Nations has partnered with Google to develop the UN System Data Commons, aimed at enhancing accessibility to UN statistics for AI applications. This initiative involves 26 UN entities and is set to launch with 20 datasets, with a goal of having 80% of UN statistical datasets available on the platform by 2027. Google.org has committed $2 million for capacity-building and technical support.

A benchmark study by UNICEF revealed that six large language models achieved an average accuracy score of only 21.2% on global development indicators, highlighting the need for improved data quality. The study gathered 133,000 responses and noted a 67% year-over-year increase in traffic to UNICEF's data website from generative AI assistants, with AI referrals accounting for 10% of visits. Currently, ChatGPT referrals contribute to 6.4% of all sessions on the site.

Shantanu Mukherjee, acting director of the UN Statistics Division, emphasized the project's scale and flexibility, stating it connects multiple UN agencies for the first time. Prem Ramaswami, lead of Google’s Data Commons team, cautioned that while AI models can assist in data interpretation, human review is essential to avoid misinterpretation of nuanced information. This collaboration follows Google's previous efforts, including the launch of Data Commons in 2018 and recent support for the Model Context Protocol (MCP) introduced last year.