MS “40.6% of Koreans use AI, ranking 12th in the world…UAE tops the list with 73.3%”

A recent survey conducted by Microsoft's AI Economy Institute shows that 40.6% of users in South Korea utilize generative artificial intelligence (AI) products. This marks a 3.5% increase from the previous quarter, bringing the country's AI user rate to 40.6% based on June data. The percentage was calculated by measuring the proportion of the production-age population (15 to 64 years old) using generative AI products. Globally, the AI user rate increased by 1% from the previous quarter to 18.8%.

South Korea ranks 12th in the world in terms of AI usage, up from 16th in the first quarter. Microsoft noted that the country has recorded the largest percentage growth among major economies for three consecutive periods, with the AI user rate rising by 15 percentage points over the past year.

At the national level, the UAE leads with an AI usage rate of 73.3%, followed by Singapore (64.3%), Ireland (49.9%), and France (49.6%). Japan, which saw a 2.2% increase from the previous quarter, recorded 24.7% in this latest period, marking the largest percentage growth of any country.

The survey reveals a growing disparity in AI usage between developed and developing economies. High-income economies, referred to as the "Global North," have an AI usage rate of 28.8%, while the "Global South," primarily consisting of emerging nations, stands at 16.2%. This 12.6 percentage point gap is higher than the previous half-year (10.6 percentage points) and first-quarter data (12.1 percentage points).

The study also indicates a surge in the use of open-weight (open-source) models globally. Microsoft's analysis of data from its AI model marketplace, OpenRouter, found that open-weight models accounted for 75% of the total token usage, a significant increase from the 25% level at the end of last year. This is more than double the previous figure.

Open-weight models allow developers to modify models according to their needs or run them in their own environments, making customization and deployment easier compared to closed models. This research was based on aggregated and anonymized user data analyzed by Microsoft.