StatGPT: AI-Ready Official Statistics

Why It Matters

StatGPT addresses a growing trust gap as users turn to AI for data: by ensuring answers come directly from authoritative sources, it protects the integrity of policy-critical statistics and helps maintain reliable inputs for economic decision-making. Broad adoption will shape how governments and markets find and use official data in an AI-first world.

Summary

The IMF today launched the public preview of StatGPT, an AI interface designed to query official statistics by translating natural-language questions into structured API calls to national and international statistical databases rather than generating numbers from models. Speakers warned that general-purpose large language models often produce plausible but incorrect numerical outputs, risking misinformation and eroding trust in official data. StatGPT aims to preserve statistical precision and authority while improving accessibility, and the IMF emphasized the need for richer metadata, robust APIs and governance to integrate official statistics into AI-driven information ecosystems. Panelists urged a collective effort among international organizations, national statistical offices, technology partners and the private sector to make official data an AI-ready global public good.

Original Description

As AI becomes increasingly integrated into economic work, trust in the underlying data matters more than ever. StatGPT demonstrates how natural language AI can provide direct access to official statistics while ensuring the accuracy and integrity of published figures.

The launch event will feature opening remarks by IMF Deputy Managing Director Bo Li, a panel discussion with speakers including IMF Chief Statistician Bert Kroese, IMF Chief Information Officer Shirin Hamid, STA Deputy Director Jim Tebrake, Bruno Tissot, Head of Statistics and Research Support at the Bank for International Settlements, and EPAM Chief Strategy and Transformation Officer Elaina Shekhter, as well as a live demonstration of StatGPT and its potential applications for economic analysis and policy work.