Voir tous les articles

Big Tech’s AI Isn’t The Future Europe Needs

As world leaders gather for the AI Action Summit in Paris today and tomorrow, we face a pivotal moment in the trajectory of artificial intelligence. The suite of technologies commonly labelled “AI” have already transformed industries and promise to reshape societies. But the critical question remains: in whose interest is this transformation occurring, and what kind of future is it building?

The biologist Theodosius Dobzhansky famously remarked, ‘Nothing makes sense in biology except in light of evolution.’ In the context of AI, we might say, ‘not much makes sense except in light of power struggles.’ These struggles dictate who controls AI, whose interests it serves, and which values guide its development. Today, the power is mostly concentrated in the hands of a few tech giants.

History teaches us the dangers of concentrated power. In medieval Europe, improvements in agriculture increased productivity but did little to improve the lives of workers. The nobility and clergy, who owned the assets and controlled wealth, enjoyed all the fruits of better technologies and agricultural organisation, while workers continued to struggle in poverty. The path that AI takes will, similarly, determine how economic gains are shared throughout the population, as well as shaping the fabric of the societies we all live in.

There are two broad directions for AI. 

The first is a relentless pursuit of artificial general intelligence (AGI) ¹ and then superintelligence, where machines outperform humans in almost every task. While such a vision may conjure fears of machines taking over, the primary threat in this scenario would come from the unchecked power of those who design and control these systems. Such a future would significantly deepen inequality. By removing our agency, it would also diminish and dilute what it means to be human. 

It remains questionable whether AGI is actually achievable in the near future. Even if it were, it is unlikely to deliver the promised productivity gains. A more likely scenario is that substandard AI systems displace workers from roles where they bring expertise and insight, thereby undermining economic value, not creating it.

The second path is what my colleagues and I refer to as “pro-worker AI” ² or “pro-human AI.” This vision sees AI as a tool to empower individuals, making workers more productive by providing context-specific, reliable information that complements their expertise. It prioritises giving people control over their own data ³ and enabling them to perform a broader range of tasks with greater confidence and capability.

This second vision is not a pipe dream. Today’s AI can already create systems that genuinely assist workers and citizens. Yet this potential will be under-realised within an architecture focused on mimicking or outperforming humans rather than supporting them. Instead of creating tools to enhance decision-making, many companies seem preoccupied with developing models that churn out hollow Shakespearean sonnets or other shallow, lifeless imitations. To preserve what makes us human (and to let Shakespeare rest peacefully), AI must move beyond imitation. It should provide clear, interpretable advice to human decision-makers, helping them make more informed choices.

The trajectory of the tech industry so far reflects deliberate choices rooted in both ideology and economics. Ideologically, the industry is motivated by dreams of AGI and superintelligence (and itself reshaping society via new powerful technologies). Economically, Big Tech has thrived on business models that generate massive profits by automating tasks ⁴ , cutting labour costs ⁵ , and monopolising digital advertising, with little interest in empowering workers or strengthening democracies. New, more socially beneficial business models can replace this status quo if new firms are given a chance to succeed. However, current market conditions make it easy for incumbents to dominate — because they have all the cash (and can acquire or bury competitors), all the data, huge existing networks of customers, and the complicity of lawmakers who have given up on antitrust.

If the world was under any illusion that Big Tech’s power would be tempered by US government regulation, images of the tech CEOs at Donald Trump’s inauguration have certainly squashed this notion. Protected and bolstered by the new US administration, the direction for Big Tech companies in their relentless pursuit of AI is clear: they plan to use the technology as a tool to entrench their dominance and reshape global markets to serve their own interests.

Yet it doesn’t have to be this way. As the relationship between the US and EU strains, the AI Action Summit represents an opportunity for Europe to course correct its AI future.

Europe cannot become a passive consumer of these systems, designed with little regard for economic sovereignty, capacity for innovation, or democratic values. The recent emergence of DeepSeek’s Large Language Model shows that innovation can still trump scale, if given a chance. 

By tackling the power and influence of these companies, for example with antitrust action, and embracing a vision of “pro-human AI”, governments can create a truly competitive environment. Only then might technology help workers and citizens, rather than becoming a tool for a very small elite to dominate the rest of humanity.

1. Daron Acemoğlu, «The Simple Macroeconomics of AI», MIT, 5 April 2024.
2. Daron Acemoğlu, David Autor, Simon Johnson, «Can we Have Pro-Worker AI? Choosing a path of machines in service of minds», Policy Memo, Shaping the Future of Work, MIT September 2023.
3. Katharine Miller, «Privacy in an AI Era: How Do We Protect Our Personal Information?», Stanford University, 18 March 2024.
4. Richie Koch, «Big Tech has already made enough money in 2024 to pay all its 2023 fines», Proton, 8 January 2024.
5. Camilla Hodgson, «Tech companies axe 34,000 jobs since start of year in pivot to AI», The Financial Times, 11 February 2024.