A survey reveals that South Korea's share of generative AI users has exceeded 40% for the first time.
According to Microsoft (MS) AI Economy Institute's 'Q2 Global AI Diffusion Report' released on the 28th, the proportion of domestic generative AI users reached 40.6% as of June, up 3.5 percentage points (p) from the first quarter. South Korea also rose four spots to 12th globally.
MS analyzed that South Korea recorded the largest quarter-on-quarter increase among major economies for three consecutive quarters. Over the past year, the domestic AI user share increased by approximately 15%p.
Globally, generative AI adoption expanded. The proportion of generative AI users among the global working-age population (15-64) rose from 17.8% in Q1 to 18.8% in Q2, a 1.0%p increase. By country, the United Arab Emirates led with 73.3%, followed by Singapore (64.3%), Ireland (49.9%), and France (49.6%).
The gap in AI accessibility between regions widened. In Q2, the 'Global North' (advanced economies) had a user share of 28.8%, 12.6%p higher than the 'Global South' (emerging economies). The gap has steadily expanded from 10.6%p in late 2023 to 12.1%p in Q1 and 12.6%p in Q2. MS cited infrastructure (power, internet), digital literacy, and local language support as key factors influencing AI diffusion.
Regional differences in AI usage patterns also emerged. In the Global South, AI was more commonly used for self-directed learning, skill development, and academic support. In the Global North, consumption purposes such as shopping, search, and feedback were relatively more prevalent.
The report also highlighted the growing adoption of 'open-weight' models, which reduce costs and localization barriers. These models publicly share trained weights, allowing developers to customize or deploy them locally. According to MS's analysis of AI broker platform 'OpenRouter' data, 75% of total token usage originated from open-weight models—more than double the one-third share recorded late last year.
The report measured AI diffusion by calculating the proportion of the global working-age population (15-64) who used generative AI products during the period, based on aggregated and anonymized usage data. The metric also factored in operating system and device market share, internet penetration rates, and population size variations across countries.
