When One-Third Isn't Enough – AI Is Widening Inequality Across Europe
Birthday poems. Cake recipes. AI memes. Europeans are using artificial intelligence, just not at work. The technology that promised to transform productivity has become a pastime instead of a tool. And whilst the Nordics integrate AI into their workplaces, southern Europe watches from behind.
Text: Martti Asikainen, 17.1.2026 | Photo: Adobe Stock Photos
Europeans are using artificial intelligence for birthday poems and cake recipes, but not for work processes. The technology that was supposed to revolutionise productivity has become, in much of Europe, a leisure-time gadget. Meanwhile, the continent’s internal divide is growing as the Nordic countries race ahead, and the south falls behind.
AI has quickly become new basic infrastructure. It’s run in the cloud, integrated into office software and customer service. At the same time, it’s merging into everyday life in ways made invisible – into mobile phones, apps, marketing platforms. You’ll find it in Alexa, Gemini, Siri, and even in your fridge, if it happens to be the latest model.
Eurostat’s recent figures confirm the shift. Last year, as many as 32.7% of Europeans used generative AI tools (Eurostat 2025a). On the surface, this looks like a boom and a technological breakthrough, but closer inspection reveals something more awkward. The technology that was meant to democratise knowledge and efficiency appears to be building a new kind of divide – and at worst, reinforcing the old one.
Leisure First, Work Second
According to Eurostat, most users deploy AI for personal purposes: 25.1% in their free time, 15.1% at work, and just 9.4% in education (Eurostat 2025a). ChatGPT writes birthday poems but doesn’t bend to work processes. Midjourney creates stunning images, but somehow isn’t used for communications. The technology that was supposed to transform how we work functions mainly as a leisure tool – like sports equipment that’s never taken onto the pitch.
At the same time, something else is off-kilter. Historically, new technologies – the computer, email, the smartphone – spread from work to leisure and everyday life. Now the pattern is running in completely the opposite direction. AI is spreading from leisure into work. The reason is probably its ease of use. Leisure use doesn’t require change management, guidance, data protection, procurement processes, or risk management. Curiosity and an internet connection are enough.
In the workplace, AI immediately collides with questions like: who’s responsible for the outcome? What data can be fed in? How are errors spotted? And what happens if the model hallucinates its answer? European organisations haven’t yet built the systems, training, and processes that would make AI use natural and safe.
In schools, the situation is even more challenging. Only about one in ten uses AI for studying (Eurostat 2025). Here it’s not just about willingness, but about constraints: GDPR, protection of minors, approved tools, guidance, and pedagogical responsibility. European education providers’ guidelines emphasise that the use of generative AI requires a legally and pedagogically sound framework (EURSC 2025).
The Skills Gap Is a Real Problem
Colin van Noorden, a researcher at KU Leuven in Belgium, summed up the core of the problem beautifully, I think: people across Europe say they don’t know what they could use AI for (Yanatma 2025). This reveals the true nature of the modern digital divide. It’s not about access to technology, but about the ability and skill to use it meaningfully.
We all have smartphones bursting with intelligence in our pockets, but few can build a business with one. Likewise, almost everyone has access to ChatGPT or other large language models, but few integrate it into their work processes. AI doesn’t reward those who have the app, but those who have the knowledge, the process, and permission to use it at work (OECD 2025a).
The European Commission has recognised that AI competence, skills, and literacy are central factors in achieving the Digital Decade targets (European Commission 2025a). The more openly the subject is discussed, and the more actively organisations share their successes and practical examples, the more certain it becomes that organisational support will improve and AI will shift from experiment to work process.
A 30 Percent Gap Divides Europe
The disparities within Europe are stark. Norway is the clear leader in AI use with an adoption rate of 56.3%. Denmark is second with 48.4%, and Switzerland third with 47.0% (Eurostat 2025a). The gaps at the bottom are widening dramatically. At the other end of the statistics are Romania at 17.8%, Italy at 19.9%, and Bulgaria at 22.5% (ibid. 2025a). We’re looking at a gap of over 30 percentage points within the EU alone.
The problem, as I see it, isn’t purely about accessibility. Many of the applications are free and work on virtually any device. The gap is rather one of capability and organisational culture. Evidence of this comes from the differences between the Netherlands and Greece. In the Netherlands, personal and work use of generative AI are almost balanced (28% in leisure time, 27% at work), whilst in Greece the corresponding figures are 41% and 16%.
A gap of nearly 25 percentage points reveals that in the Netherlands, AI is integrated into both spheres of life, whilst in Greece it’s mainly a leisure phenomenon. Not because Greeks are less interested in using AI at work, but because workplaces lack the infrastructure, training, and culture to harness it (e.g. OECD 2025b; European Commission 2025b). A harsh lesson the EU must learn time and again: equal access doesn’t produce equal outcomes without equalising support.
Why This Matters Right Now
The World Economic Forum regards AI as one of the most central forces shaping labour markets between 2025 and 2030 (WEF 2025). It’s expected to transform tasks, skill requirements, and how productivity is distributed. Those countries and organisations that adopt AI now are building a lead that will be difficult to close later, because the advantage isn’t a software licence – it’s a way of working.
The International Labour Organization also emphasises that “exposure” to generative AI is uneven. It affects some professions and tasks far more than others (Gmyrek et al. 2025). For some, AI is a toolkit that boosts productivity. For others, it’s background noise that doesn’t touch their work at all. And for a third group, it may be a threat that changes working conditions without any chance to influence it.
The European Parliament’s research services have highlighted that AI’s impacts on working life aren’t merely technical, but relate to power, responsibility, and on whose terms algorithms organise work (EPRS 2025a). This is another reason why the earlier-mentioned “I don’t know what I’d use AI for” isn’t just an individual skills deficit. It’s a sign that the system doesn’t yet support new ways of working.
What the Statistics Don't Reveal
Eurostat’s figures don’t tell us about the quality of use. We can’t deduce from them whether it’s trivial tasks or genuine problem-solving. Nor do we see whether use is random experimentation or systematic integration into work processes. When someone asks ChatGPT for the roasting time for a ham, nothing changes. But when an analyst asks AI to identify anomalies in thousands of rows of data, or a solicitor identifies risk points in a contract, it can have far-reaching effects on the nature of work.
What I mean by this is that the presence of technology isn’t the same thing as its impact. On the other hand, perhaps the most critical question concerns those who don’t use AI at all. Do the 67.3% left outside even have the possibility to utilise AI, or has the choice been made for them? I believe for the majority of them, the choice has been made structurally. Perhaps they work in organisations where AI is still regarded with suspicion.
Or they may live in countries where digital infrastructure and the skills base are weaker, and work in sectors where technological renewal is slower than average (European Parliament 2025b). The fact is, though, that the Nordic countries, the Baltics, and parts of Western Europe are accelerating forwards. Southern and Eastern Europe are falling behind. The gap isn’t merely technical, but also institutional and cultural.
Infrastructure Requires Investment
If AI is new basic infrastructure, we should treat it like infrastructure. It requires investment, guidance, training and development, and fair access to meaningful use as well. The EU has recognised that skills and adoption capacity will determine the success of Digital Decade objectives (European Commission 2022; European Commission 2025b). This isn’t about who can be bothered to learn, but about Europe’s competitiveness and equality.
Why does Norway lead at 56%, whilst Romania stays at 18%? And what explains why in the Netherlands AI is part of working life, but in Greece it’s viewed with suspicion? Eurostat’s 32.7% adoption rate tells us that the technology has become everyday. But it doesn’t tell us whether anything has truly changed. And that’s precisely what matters. Change. In my view, every European actor should ask themselves: are we building divides with AI, or bridges?
References
European Commission (2022). Human Capital and Digital Skills in the Digital Economy and Society Index. European Commission. Brussels.
European Commission (2025a). AI talent, skills and literacy. Published 4 December 2025. Accessed 17 January 2026.
European Commission (2025b). State of the Digital Decade 2025 report. Published 16 June 2025. Accessed 17.1.2026.
European Parliament Research Service EPRS. (2025). Digitalisation, artificial intelligence and algorithmic management in the workplace: Shaping the future of work. European Parliament. Brussels.
EURSC. Office of the Secretary-General of the European Schools. (2025). Legal and pedagogical guidelines for the educational use of generative artificial intelligence in the European Schools. Brussels.
Eurostat. (2025). 32.7% of EU people used generative AI tools in 2025. Published on the European Union website 16 December 2025. Accessed 27 December 2025.
OECD (2025a), Generative AI and the SME Workforce: New Survey Evidence. Published 5 November 2025. OECD Publishing. Paris.
OECD. (2025b). Bridging the AI skills gap: Is training keeping up? Published 24 April 2025. OECD Publishing. Paris.
Gmyrek, P., Berg, J., Kamiński, K., Konopczyński, F., Ładna, A., Nafradi, B., Rosłaniec, K. & Troszyński, M. (2025). Generative AI and Jobs. A Refined Global Index of Occupational Exposure. ILO working paper, 140. International Labour Organization (ILO). Published May 2025.
Yanatma, S. (2025). ChatGPT, Gemini, Grok and others: Which countries use generative AI tools most across Europe? Published in Euronews 28.12.2025. Accessed 17.1.2026.
World Economic Forum. (2025). Future of Jobs Report. Insight Report. Published January 2025. Cologne & Geneva.
Martti Asikainen
Communications Lead
Finnish AI Region
+358 44 920 7374
martti.asikainen@haaga-helia.fi
Finnish AI Region
2022-2025.
Media contacts