Unpacking how the world is using AI | Microsoft's AI Diffusion Report

Key Concepts

• AI Diffusion: The rate and scale at which AI technology is adopted across different populations and geographies.
• Working-Age Population: Defined by the OECD as individuals between 15 and 65 years old.
• AI Divide: The disparity in AI adoption rates between different regions (Global North vs. Global South) and demographics (Urban vs. Rural).
• Agentic AI: AI systems capable of performing tasks and automating processes, often through natural language interaction rather than traditional programming.
• GitHub Copilot: A tool that enables users to generate code using natural language (e.g., English) instead of traditional programming languages like Python or C#.
• Elasticity of Demand: An economic concept suggesting that as productivity increases and costs decrease, the demand for the output (software) grows, preventing job displacement.

---

1. Global AI Diffusion Trends

Juan Lavista Ferres, Director of Microsoft’s AI for Good Lab, reports that AI adoption is accelerating globally.

• Current Statistics: As of March 2026, 17.8% of the global working-age population is using AI, a 1.5% increase from December. If this trend continues, annual growth is projected at 6%.
• Regional Leaders: The UAE leads the world with over 70% of its working-age population using AI.
• Methodology: Microsoft utilizes data from billions of devices, adjusted for internet penetration and device type (mobile vs. PC), to ensure an "apples-to-apples" comparison across countries and counties.

2. The Role of Language in Adoption

A primary driver for the recent surge in AI adoption in Asia (specifically Japan, Thailand, and South Korea) is the improvement in linguistic capabilities.

• Performance Parity: Models that previously struggled with non-English languages have improved significantly. AI performance in Japanese and Korean is now nearly on par with English, making the tools genuinely useful for local populations.
• Data Availability: The "language gap" was historically caused by a shortage of training data in non-English languages. Increased investment by companies like OpenAI and Anthropic has bridged this gap, leading to higher adoption rates.

3. Real-World Application: Thailand’s OECD Case Study

The report highlights a significant efficiency gain in Thailand:

• The Challenge: Thailand needed to translate 70,000 laws into English and compare them against 270 OECD standards to apply for membership—a process typically taking 3–5 years.
• The Solution: A team of five government lawyers used AI to perform the translation and comparison in just three months.
• Takeaway: AI eliminates "drudgery," allowing humans to focus on high-level decision-making rather than manual, labor-intensive tasks.

4. The AI Divide: Global and Domestic

The report identifies persistent disparities that hinder equitable AI growth:

• Global North vs. South: The Global North is growing at 2.8 points per quarter, while the Global South grows at 1.3 points. This is driven by foundational gaps in electricity (98% vs 88.9%), internet access (90% vs 65.7%), and digital skills (70% vs 48.2%).
• US Urban vs. Rural: Within the US, rural counties show roughly half the AI adoption of urban areas.
• The Trust Factor: There is a strong inverse correlation between trust and AI adoption. Rural areas report lower trust in AI, which may be a significant barrier to entry despite the potential benefits for rural sectors like healthcare and agriculture.

5. The Future of Software Development

Lavista Ferres argues that the definition of a "software developer" is evolving.

• Coding in English: The ability to communicate with computers to automate processes is becoming more important than knowing a specific syntax like Python.
• Economic Perspective: Contrary to the fear that AI will reduce the need for developers, history suggests that increased productivity leads to higher demand. Just as the washing machine tripled the frequency of laundry, AI makes software cheaper to produce, leading to an explosion in the creation and use of software.
• Democratization: Similar to how Windows democratized PC usage in the 80s and 90s, AI is turning coding into a universal tool—the "new Excel"—that will be used by lawyers, accountants, and architects to enhance their professional output.

Synthesis and Conclusion

The AI for Good Lab’s report confirms that we are in a period of rapid, accelerating AI diffusion. While the technology is proving to be a powerful equalizer—enabling non-native English speakers to code and governments to streamline complex legal processes—the "AI divide" remains a critical challenge. Closing this gap requires more than just software; it requires addressing fundamental infrastructure deficits in electricity, internet access, and digital literacy. Ultimately, the shift toward natural language interaction with computers suggests that the future of work will involve a broader segment of the population becoming "developers," using AI to solve problems that were previously impossible to address.