Digital Economy

Artificial intelligence grows but creates more and more inequality

The new AI Diffusion Report reveals that one in six people worldwide use generative AI, but the distribution remains dramatically uneven. UAE tops with 64%, while Africa discovers DeepSeek

by Marco Trabucchi

(Adobe Stock)

4' min read

Translated by AI
Versione italiana

4' min read

Translated by AI
Versione italiana

One in six people worldwide now uses generative artificial intelligence tools. This might sound like an achievement, but Microsoft's new report on AI deployment for the second half of 2025 tells a more complex story: while global adoption rises to 16.3 per cent (up 1.2 per cent from the first half of 2025), the gap between developed countries and the rest of the world continues to widen.

The gap between North and South

The numbers capture an accurate reality: in the Global North (developed countries) adoption has grown almost twice as fast as in the Global South (developing countries). In the second half of 2025, 24.7 per cent of the working-age population in rich countries use generative AI tools, compared to just 14.1 per cent in the rest of the world. The gap has increased from 9.8 to 10.6 percentage points in six months. To put it crudely: while Silicon Valley debates yet another language model, half the planet still struggles to access a stable internet.All ten countries with the fastest growth in AI adoption are high-income economies.

Loading...

The Emirates and the US paradox

At the top of the global ranking is the United Arab Emirates with a 64% adoption rate among the working-age population, up from 59.4% at the beginning of the year. This record is no accident: the Emirates appointed the first Minister of State for Artificial Intelligence in 2017, five years before ChatGPT became a mass phenomenon. While other governments were still debating whether AI deserved political attention, Abu Dhabi had already launched a national strategy covering nine priority areas. The result is a 67% level of trust in AI according to the Edelman Trust Barometer 2025, compared to 32% in the US and a similar European average.

Singapore retains the second position with 60.9% adoption (up from 58.6% in the first half of the year), followed by Norway in third place with 46.4% (up from 45.3%) and Ireland in fourth with 44.6% (up from 41.7%). France, Spain, New Zealand, the Netherlands, the United Kingdom and Qatar complete the top ten. Italy ranks 26th with 27.8% adoption, up from 25.8% in the first half of the year, broadly in line with the European average but far behind the Scandinavian and Gulf leaders. The United States, despite its technological leadership, is only 24th with 28.3%, even down from the previous half-year. The paradox is obvious: the country that develops the most advanced models struggles to translate this excellence into widespread use among its citizens.

The Microsoft report here highlights a crucial point: leadership in innovation and cutting-edge infrastructure alone do not guarantee widespread adoption. The US dominates in frontier modelling and cloud infrastructure, but lags behind smaller, highly digitised economies with focused national AI strategies. The trust gap revealed by the Edelman Trust Barometer suggests that concerns about privacy, misinformation and impact on work weigh more heavily than elsewhere.

South Korea: when politics and technology meet

The most interesting case of the year is South Korea, which jumped from 25th to 18th place in the ranking with an increase of almost 5 percentage points. The 81.4 per cent growth in the user base from the first to the second half of 2025 far exceeds the global average (35 per cent) and the US average (25 per cent). It is no coincidence that Seoul is now the world's second largest market for ChatGPT subscribers, just behind the US. Among the many factors that have triggered this surge, the report explains, are. Starting, again, with forward-looking national policies. In September, the government reconstituted the National AI Strategy Committee and passed the AI Basic Act, legislation that balances innovation and governance. Not recommendations, but formal mechanisms to coordinate infrastructure, regulation and public projects.

But among the factors driving the massive use of AI are also cultural factors. On the one hand, there is the dramatic improvement in the performance of language models in Korean. GPT-3.5 scored 16 points on the Korean University Admission Test (CSAT), GPT-4o rose to 75, GPT-5 hit 100. In practice, it went from below adult reading proficiency to outstanding university student performance. And then, ultimately, AI was a viral cultural phenomenon. In April 2025, the Studio Ghibli-style images generated by ChatGPT-4o popularized Korean social media. Millions of people used AI for the first time and many continued to explore its capabilities even after the trend died down.

DeepSeek: the Chinese outsider conquering Africa

The real surprise of 2025 is called DeepSeek, an open-source Chinese AI platform that has gained ground in markets traditionally ignored by Western providers. By releasing the model under an MIT licence and offering a completely free chatbot, DeepSeek has removed financial and technical barriers that limit access to advanced AI.

The results are obvious: dominant market share in China (89%), but also in Russia (43%), Belarus (56%), Cuba (49%) and Iran. Even more significant is the explosion in Africa, where DeepSeek usage is estimated to be between 2 and 4 times higher than in other regions. Credit is also due to strategic promotions and partnerships with companies such as Huawei, which have actively deployed the platform through telecommunication services.

Copyright reserved ©
Loading...

Brand connect

Loading...

Newsletter

Notizie e approfondimenti sugli avvenimenti politici, economici e finanziari.

Iscriviti