Europeans Are More Pragmatic About AI Than Their Prophets

The crowded market of visions for artificial intelligence (AI) has gained two striking new entries in recent months. The first, courtesy of Pope Leo, draws attention to the challenge of preserving human dignity in the age of all-knowing machines. Then there is Elon Musk. In his version, AI will deliver nirvana within five to 10 years as it becomes smarter than humans, work becomes optional, and money disappears. We will become pet labradors to a superintelligence that makes every meaningful decision for us.

With AI discourse dominated by ever wilder claims, it is worth looking at the snapshot captured by Mandate’s nine-country survey. The global pollster asked nearly 14,000 respondents in the United States, the United Kingdom, and seven European countries about their attitudes towards AI.

A geographical divide, not an ideological one

The findings suggest a striking geographical divide. While 50 per cent of Americans and 46 per cent of Britons say AI’s effects on society are already negative, the figure reaches only 35 per cent across continental Europe.

Even where ideology enters the picture, geography remains decisive. Among progressive voters in the United States and the United Kingdom, majorities of as much as 57 per cent see AI as harmful. Among comparable political groups in Hungary, Poland, and Spain, however, positive and neutral answers dominate. Hostility towards AI on the left is, for now, not a European product but a distinctly Anglo-American one.

Why the difference? One reason is that Europe is not leading the frontier race. Local opposition to gigantic data centres has not gripped the continent as it has the United States. Some 63 per cent of Americans say a data centre next door would be bad news, compared with between half and a third of that figure in European countries.

Meanwhile, the wave of job losses attributed to AI has not yet begun in Europe. A technology whose costs are still largely hypothetical may be easier to embrace than one already being felt.

Usage data point in the same direction. Anglophones use AI the least: only 37 per cent of Britons and 47 per cent of Americans use an AI chatbot weekly, compared with 65 per cent of Spaniards. Given how loudly AI dominates British and American media and politics, that gap looks less like ignorance than scepticism.

But lower levels of scepticism do not necessarily amount to enthusiasm. In much of continental Europe, large numbers remain neutral or undecided about the merits of AI: 45 per cent of respondents in Hungary and Italy, 44 per cent in France, 43 per cent in Poland, and 41 per cent in Germany give a neutral or “don’t know” answer. The European contrast, then, is not simply between optimism and pessimism, but between publics that have already formed a view and those that have not.

The second major finding concerns gender. In every country surveyed, men are more positive about AI and women more negative. The technology’s dark side may offer part of the explanation: women are overwhelmingly the victims of deepfakes, non-consensual pornography, and AI-fuelled harassment.

A constructive mandate, not a backlash

Writing from Silicon Valley last month, the Guardian columnist Timothy Garton Ash asked whether even an AI disaster on the scale of Hiroshima would be enough to drive humankind to protect itself. He concluded that it probably would not. Garton Ash looks for the missing brakes on catastrophe in elite coordination, treaties, and elsewhere. Yet he fails to consider the publics themselves, and their role or leverage.

Mandate’s survey suggests a more mundane, and perhaps more useful, picture. Public opinion is ahead of the discourse on AI. People neither long for labrador-hood nor await the apocalypse with folded arms. They use the technology, see its benefits, and demand rules, responsibility, and protection: what one would expect of an adult society.

What differs between the Anglosphere and continental Europe, when it comes to rules, is the starting point. In the former, the demand is for restraint of a technology already judged harmful. In the latter, it is for guardrails around a technology that people have not yet fully decided how to judge.

That is a constructive mandate, not a backlash. The real gap in AI governance is not between reckless publics and responsible elites. It is between publics that already want regulation and institutions that have yet to turn this demand into policy.

Fear alone will not bridge that gap. No technology in history has spread this fast, and people do not embrace something at that pace out of naivety. They see its uses. A total ban would be easy to demand and bad politics to run on, the kind of policy that loses to lived reality within an election cycle. Even Pope Leo calls for rules, not prohibition.

AI therefore presents both an opportunity and a test for Europe. The demand for serious guardrails is already there. The hard part is what follows: writing good rules, then doing the still harder work of enforcing them.

For European progressives, the warning is equally clear. Do not import the Anglo-American culture war over AI, in which the left owns the fear and the right owns the future. The position already visible in the data is simpler: embrace the technology, protect workers, and write the rules.

AI remains a low-salience but high-intensity issue. You cannot build a coalition on it, but you can lose one with it all too easily. Europe has a window before its publics make up their minds. Better, then, to listen to them before the bill arrives. For once, they are ahead of their prophets.

A fuller summary of the Mandate research is available (in Hungarian) on the website of the Digital Sociology Research Centre.