On 1 and 2 November 2023, politicians, scientists, and leaders of tech companies gathered in Britain's Bletchley Park (Milmo and Stacey 2023) to discuss the risks of artificial intelligence (AI). The main focus points were disinformation, cyber attacks, and the (by the way probably low) likelihood that AI will soon go beyond a “frontier” and threaten life as we know it.

As often happens on such occasions, the real issues with AI remained undiscussed: the power to control this promising technology is mainly in the hands of a few very large commercial giants (Maslej et al. 2023), in some of which only a small number of CEOs and major shareholders call the shots (LeFrance 2021). As property owners of AI and the necessary infrastructure, these companies almost unilaterally decide what AI applications will be developed, at what pace, the acceptable risks in the process, the values AI should serve, and under what conditions others may use AI. In doing so, their own private interests (or delusions) are usually paramount. Not so much the public interest.

How did we end up in this “feudal” (Kotkin 2020; Varoufakis 2023) situation?

To answer this question, we need to look first not at the technology itself, but at the four dimensions of societal life (that is, the political, legal, economic and financial, and sociocultural and socioenvironmental dimensions) from which it emerged. Perhaps the most defining institution for AI is the modern system of property rights. “Invented” (Blaufarb 2016) over two centuries ago, it introduced the possibility for individuals, as “property owners,” to concentrate unlimited amounts of exclusive possession of technology and exploit it as they saw fit—no matter how important that technology was to the rest of society.

Then, during the course of the nineteenth century, legislatures and other public authorities developed the notion of the private corporation—the “joint-stock company” in legal terms (Micklethwait and Wooldridge 2003). Among other things, this turned companies into independent “legal persons” and allowed them to issue shares. Indeed, this was perhaps the most consequential innovation of modern times. It further stimulated the concentration of property ownership of, and thus control over, technology in the hands of a few. Soon, around 1900, mega-corporations became the dominant technological players in the West—think of oil, cars, medicine, food, and weapons.

In theory, these mega-corporations could of course be “regulated” by legislatures and other public authorities. But in practice, that was easier said than done due to the emergence of disproportionate lobbies, dependencies linked to jobs and war, and so-called “liberal” ideologies.

The globalization of the world economy throughout the twentieth century saw mega-owners of technology become very large “multinationals” that proved even harder to tame. Thanks to international private law (Pistor 2020), multinationals could move their operating units from country to country. By doing so, they could avoid legislation or gain more favorable arrangements by playing countries off against each other. What followed was a race to the bottom between countries—in which democracy, the environment, and the world's weakest lost out and multinationals become major, semi-independent political players (Mikler 2018). Technological “progress,” therefore, has not exactly worked for everyone!

With the arrival of digitalization, the race to the bottom between countries and the consolidation of technological power and wealth among a few mega-corporations gained momentum. Those who already own certain technology can develop digital technologies further. Those who develop digital technologies further have more users, data, and thus opportunities to add to it. Whoever has the best digital technologies then, in turn, attracts more users, can collect more data, and so it goes on. With such strong “network effects,” it is not only a case of the rich getting richer (and the poor getting poorer), but also “the winner takes all.”

Digital technologies have made mega-corporations ever more mobile, taking them even further afield than before (even into space, like Elon Musk's Starlink). In this way, they can evade state, federal, and EU legislation even more easily than previous generations of mega-corporations and be more effective in playing countries off against each other to achieve a more favorable business climate for their own operations. Think of even lower taxes, more flexible labor law (or less enforcement of it), lower energy costs, and so on. It is not without reason that Big Tech companies—the very largest and richest companies on earth—get huge subsidies and other benefits from states and supranational governments to establish themselves in their territories (Larson 2020). States, competing against each other, are forced to pull out all the stops to make sure they get another piece of the economic pie.

The digital transition has also increased society's dependency on mega-corporations. Even governments are now moving to the Big Tech cloud, using one Big Tech email server, one Big Tech operating system, and so on. Moreover, as the new “system technology” (Sheikh et al. 2023), the digital technologies of mega-corporations are interweaving with other types of technology and the social fabric of society. This does not make it any easier for states and their international alliance to protect the public interest.

Meanwhile, power relations in the technology sector have become so distorted that “tech regulation” can be counterproductive (The Economist 2023). After all, mega-corporations—helped by the most expensive lawyers—have the means to evade new rules or bend them to their will. Smaller companies and startups are unable to do this. Mega-corporations also have much greater means to work around increasingly complex legal environments than their (potential) competitors.

In short, the real problem of AI has less do to with AI itself than most leaders at Bletchley Park seem to think, let alone with the AI “mass extinction” with which Big Tech and its allies have tried to distract us once more from the real issues surrounding AI. Far more acute is the broad social inequality that today's AI reflects and further exacerbates. If that is not overcome, we will, at best, be fighting a losing battle. It is more likely, though, that the specific AI laws now being proposed by politicians will only lead to a further concentration of technological power in the hands of a few private giants—with all the associated negative consequences and risks.

AI demands a different legal system: limited and less absolute property rights (Robeyns 2024), more responsible corporate laws (Pistor 2020; Robé 2020), and a private international law that works to the benefit of ordinary people, constitutionalism, and democracy (Rodick 2011). That, however, is a long-term project. In the short term, it is crucial that countries reduce their dependence on the very biggest tech companies. Governments should take the lead. Only then can they (once again) be capable of safeguarding the public interest, including such fundamental values as democracy, the rule of law, and constitutionalism.

As long as governments and other important societal actors continue to make themselves more dependent on a few commercial mega-owners—and welcome them as heroes as seen at summits like Bletchley Park—AI cannot possibly work for all of us.

Translation: Wendy H. Rodger