Access to capital and rules, so Europe can meet the challenge on Ai

"Europe must find its own way to support innovation": Daron Acemoglu, winner of the 2024 Nobel Prize in economics and professor at the MIT in Boston (USA) is the economist who has most lucidly and courageously broken through the techno-optimism of the last ten years. He has shown that the equivalence between progress and growth in recent decades must be rewritten. More. Over the past 40 years, new technological advances have increased inequality and have not led to an increase in productivity. And it is precisely 'productivity' that was the word he focused on most here at the Trento Festival of Economics, where he spoke via a link. Now, with artificial intelligence, we have an opportunity to reverse this trend. But the decisions taken by companies and governments today will determine whether AI will bring a more widely shared prosperity or simply reinforce the current trend. The need,' he explained during the lecture, 'is to put in place coherent regulations at European level that apply to all companies, rather than fragmented national rules. This would simplify operations for companies across the continent.

It was precisely on Europe that he focused most. Since the appearance of DeepSeek, the very powerful Chinese chatbot, a 'Sputnik' effect has been triggered, with China and the US engaged in a race reminiscent of the space race, which confirms the mantra: the US invents, China copies, Europe dictates the rules. Even in the age of artificial intelligence. "Europe should find its own way to provide 'cheap credit' for innovations, as distinct from venture capital, to enable scalability. This is also crucial for competition, as the lack of growing companies hampers competition'. The continent of Europe is in some ways the focus of the Nobel Prize winner's hopes. Size gives us a special place and a 'technical' advantage, especially for having set a different AI direction from China and the US, one that is more respectful of human rights, controllable and explainable. Acemoglu's suggestion is to pit its 'competitive power' against that of China and the US. He also proposes that other emerging economies, such as Saudi Arabia, Brazil, Mexico and India, could join forces with Europe or form a 'fourth bloc' to support their specific voices regarding AI development and labour.

"What is needed," he emphasised, "are stronger competition policies, including the application of antitrust laws to AI, and the promotion of an 'alternative path' for AI development from Europe. Allowing uncontrolled corporate concentrations', he said, referring to Europe's initial positions on Google, 'does not help the creation of a healthy competition system'.

Acemoglu's position on the labour market also has the insight to keep the new forms of artificial intelligence on the market in the 'equation'. After chatbots, since the beginning of the year we have entered the era of autonomous agents, AI models capable of making decisions autonomously. Although there is much uncertainty as to when models will be able to interact autonomously with digital platforms, other AI tools and even humans, there is no doubt that this development will be transformative, for better or worse. Yet despite all the commentary (and hype) about 'agent-like' AI, many big questions remain unanswered, the biggest of which is: what kind of agent AI is the tech industry trying to develop? Different models will have very different implications.

With an 'AI as advisor' approach, AI agents would offer individually calibrated recommendations to human decision-makers, leaving humans always in the driving seat. But with an 'autonomous AI' model, agents would take control on behalf of humans. 'This is a distinction,' he emphasised, 'with profound and far-reaching implications'. According to Acemoglu, the potential for agents to provide real-time, customised information to individual workers for training, operating new equipment, or troubleshooting should be studied and explored. However, he warns that AI agents should not make all decisions, as this could be 'toxic' and deprive humans of their power. The ideal scenario is for AI agents to be 'at our service', not for them to become 'our vassals'.