Why the goal of AI sovereignty remains so elusive

Sam Altman meets President Sheikh Mohamed in Abu Dhabi. OpenAI says it is offering data residency services to its business customers in the UAE. UAE Presidential Court

In my experience as a technology journalist, I’ve noticed that sometimes it’s the seemingly smaller announcements that speak volumes about the status quo and what to soon expect from the sector.

The feeling was bolstered by a recent development concerning OpenAI, which helped to put AI on the map with its ChatGPT tool in late 2022, and the UAE, which has made no secret about its ambitions to be a leader in AI development.

OpenAI said it was offering data residency services to its business customers in the UAE. The company, rapidly ascending to unprecedented heights of technology influence, said that its residency services will allow for businesses to store their data in the UAE, which many say is a step in the right direction of fulfilling the ideas of data sovereignty.

Sam Altman, chief executive of OpenAI, at a the Federal Reserve Integrated Review of the Capital Framework for Large Banks Conference in Washington, DC, US, on Tuesday, July 22. In recent years, AI has given us all the tools that rival the human mind in terms of problem-solving and critical thinking. Bloomberg via Getty Images

As is the case with most developments in technology, data sovereignty can take on a few definitions depending on who you are. But most reasonable minds would agree that it can broadly be explained as a way for countries to ensure they can enable the use of computer tools and access to data without having to rely on external factors, companies or geopolitical elements beyond their control.

It’s important to point out that OpenAI is hardly the first company to talk about the idea of data sovereignty.

Several years ago, Nvidia chief executive Jensen Huang spoke about it in the context of AI sovereignty, and later also touched upon data sovereignty as the AI sector matured at a breakneck pace. Microsoft’s chief, Satya Nadella, has also spoken about it at length as the company builds its data centres around the world, often boasting of “sovereign cloud capabilities”.

Amid this corporate push for AI, tech, data and cloud sovereignty, it would be easy to chalk up the discussion to just a polite way for these companies to sell more of their products and services. It’s naive to think that’s not the case to some extent, but it’s also a gross oversimplification of how the technology sector has evolved over the years, and how we as consumers have ultimately factored in sovereignty in most of our technology purchases.

In many ways, sovereignty has been at the heart of the earliest days of personal computing.

During an interview in 1974 with an Australian reporter who was accompanied by his young son, the writer, futurist and television personality Arthur C Clarke, standing near a mainframe computer taking up an entire room, spoke about how computer terminals would some day give sovereignty to people all over the world.

“When [your son] grows up,” Clarke said referring to the reporter’s son, “he will have in his own house at least a console to which he can talk to his friendly local computer, where he can get all of the information he needs for his everyday life such as bank statements, theatre reservations ... he’ll talk to the computer and get information from it, and he’ll take it for granted.”

It’s difficult to express just how radical Clarke’s vision was at the time, and ultimately how accurate it all became. That interview came in the middle of a very much under-the-radar debate in the nascent computer sector.

Would the masses ever demand access to the massive computers that were only used by a select few? And, if so, how would they access those huge computers? Clarke’s answer mentioned the idea of consoles in every home, allowing for access to larger, off-sight mainframe computers.

He then explained that those consoles would guarantee us sovereignty by “making it possible to live anywhere we like ... any businessperson or executive can live anywhere on Earth and still do business. We won’t be stuck in cities and can live out in the country or anywhere we please”.

The invention of the personal computer just a few years later, with the introduction of the Apple II and eventually the IBM PC, took Clarke’s vision one step further. These weren’t simply consoles that provided access to bigger mainframes. They were computers in and of themselves with storage devices that allowed for the users to digitise all data, create new software and even connect with others through dial-up modems.

In the grand scheme of things, they delivered all the sovereignty they promised, along with a surge of wealth and entrepreneurial innovation that continues to this very day. What was once available to a select few through mainframe computers was bestowed upon us all.

The AI revolution, in some ways, presents us with the same challenges of mainframe computers

The internet and later high-speed internet connections made computers, laptops and then handheld devices all the more enticing, promising us unprecedented sovereignty in terms of access to information all over the world. Yet amid all that excitement, the idea of sovereignty took a back seat to the idea of connections.

We could never fit all the information we sourced from the internet on the hard drives of our personal computers. So eventually we accepted the idea of what would later be known as “the cloud”, which made it easy for us to lessen our dependence on hard drives.

In recent years, AI has caused yet another shift, giving us all the tools that rival the human mind in terms of problem-solving and critical thinking. Those tools, however, are very much dependent on data centres spread out in various parts of the world, packed to the brim with powerful semiconductors and graphics processing units. They’re also sometimes contingent on a handful of cloud computing companies that have shown that they’re far from perfect, occasionally going on the fritz and rendering millions of AI apps temporarily useless.

In turn, it’s easy to understand the appeal and persistent push for digital and data sovereignty. But that, too, pales in comparison to the ideas of empowerment first presented in the infancy of personal computing.

The AI revolution, in some ways, presents us with the same challenges of mainframe computers, in that only countries that can afford to develop AI infrastructure can take part in the promised sovereignty, while users are subliminally dependent on data centres they will probably never see.

The push for sovereignty, both in the context of data and AI overall, will truly be achieved when all that processing power is put in the hands of average consumers everywhere. As any technology expert will tell you, we’re decades away from that, but history has taught us it’s foolish to bet against the idea of progress.

That brings us back to the offering of data residency from OpenAI. It might seem like a small thing, but it’s an ambitious step in the right direction, and one of many steps needed to empower us all.