microsoft/ai-diffusion-report

microsoft/ai-diffusion-report

Description: Data and documentation for the Microsoft AI for Good Lab's AI Diffusion Reports

License: MIT

Stars: 15

Forks: 2

Open issues: 0

Created: 2026-04-30T16:47:59Z

Pushed: 2026-06-04T15:36:56Z

Default branch: main

Fork: no

Archived: no

README:

AI Diffusion Report Data

This repository contains the datasets and report materials for Microsoft’s AI Diffusion research, produced by the AI Economy Institute (AIEI).

The AI Diffusion report tracks adoption of generative AI, measuring the share of people who have used AI tools across countries, over time, and across U.S. geographies.

Overview

AI diffusion is defined as the share of people who have used a generative AI product during a given period. The measure is derived from aggregated and anonymized telemetry and adjusted to account for differences in:

• Device and operating system market share
• Internet penetration
• Country population

This repository provides the country-level datasets used in the published reports, along with archived report PDFs.

It also includes U.S. subnational AI User Share estimates at the county, state, and metropolitan statistical area (MSA) levels. These estimates use a related but distinct methodology because local estimates must account for small county samples, IP-geolocation noise, and differences in local digital infrastructure. The U.S. pipeline combines anonymized, aggregated Microsoft telemetry with a Bayesian spatial small-area model that borrows information across neighboring counties, state effects, and demographic covariates, then aggregates county estimates using working-age population weights.

Citation

To cite this work and for full methodological details, see the technical papers:

• Misra, A., Wang, J., McCullers, S., White, K., & Lavista Ferres, J. (2025). *Measuring AI Diffusion: A Population-Normalized Metric for Tracking Global AI Usage*. arXiv. https://doi.org/10.48550/arXiv.2511.02781
• Misra, A., et al. (2026). *Measuring AI Diffusion Across U.S. Geographies: County, State, and Metro Estimates*. https://aka.ms/AI_Diffusion_US_Technical_Report

Global technical report also available at: https://aka.ms/AI_Diffusion_Technical_Report

Repository Structure

data/

Contains AI diffusion datasets used in the reports, including the global country-level data and U.S. subnational data.

Typical fields include:

• country or geography identifier
• AI User Share for various periods or geographies

The data/US/ folder contains Q1 2026 county, state, and MSA estimates.

reports/

Contains PDF versions of each published AI Diffusion report.

Data Updates

Data is updated periodically alongside new AI Diffusion report releases (typically every 3-6 months).

Each update may include:

• revised country-level estimates
• additional countries or regions
• methodological refinements

Usage

The data in this repository is intended for:

• research and analysis
• policy discussions
• visualization and reporting

Limitations

AI diffusion is an evolving metric and should be interpreted with care:

• It reflects usage, not capability or impact
• Cross-country comparisons depend on adjustments for infrastructure differences
• U.S. subnational estimates are modeled small-area estimates, not direct raw county telemetry
• Estimates are subject to revision as methodology improves

No single metric fully captures global AI adoption. This dataset should be considered alongside complementary indicators where possible.

About

This work is produced by the Microsoft AI Economy Institute (AIEI) as part of ongoing research into global AI adoption and its economic and societal impacts.

For more information:

• https://www.microsoft.com/research/group/aiei/ai-diffusion/

License

MIT License