Digital sovereignty requires direction, trust and cooperation

Organisations and governments no longer seek only technological innovation, but above all control, freedom of choice and resilience. That was the take-away message of a roundtable discussion on the application of artificial intelligence (AI), organised by technology company IBM that took place in Amsterdam on Tuesday.

Digital sovereignty is now a central issue in the debate on the future of technology and AI. It has always been important, but as a result of the rapidly changing geopolitical landscape, it has become even more crucial.

According to the participants in the debate, digital sovereignty is fundamentally about the ability to use technology on one's own terms, be it data and algorithms or cloud infrastructure.

Complete independence is not even a requirement for this, but the ability to take strategic direction for oneself. Freedom of choice, or 'optionality', is crucial: organisations must be able to switch suppliers, technologies and infrastructures, without being stuck in closed ecosystems. 'Sovereignty is that optionality of the services, the data you use,' said Frank Ketelaars, senior engineer at IBM.

International chains

Complete control over the entire technology is not feasible in practice. After all, modern IT environments are intertwined with international chains of suppliers, infrastructure and knowledge, the discussion participants noted. 'If you want to be completely sovereign as a company, it is actually not possible,' Ketelaars said.

Willem Jonker, chairman of the board of the AI Coalition for the Netherlands (AIC4NL), which is working for a fast and responsible AI transition in the Netherlands, agreed. 'Control your own destiny may be a good principle, but to be able to determine your own future, you don't need to own or fully control everything yourself,' he said.

Key to data

Technological solutions can help. For instance, there are encryption techniques where organisations keep the key to their data themselves, even when that data is in the cloud - inside or outside the Netherlands. Open standards and open data formats also play a role, as they facilitate switching between systems.

Jonker emphasised not only responsible handling of AI and sovereignty, but also resilience, as an organisation and as a society. 'Those three aspects are strongly interrelated. And what we need to avoid is a Bermuda Triangle where AI disappears and we can't do anything with it anymore.'

Innovation or autonomy?

A recurring theme is the tension between sovereignty and innovation. After all, innovation thrives best in a climate without too much regulation and control. Jonker nuanced that statement. 'I don't really believe in a contradiction between sovereignty and innovation,' he said.

In his view, regulation and sovereignty need not inhibit innovation, but can actually create new market opportunities. For instance, European regulations if used strategically can lead to new industries and technological breakthroughs.

Meanwhile, Europe is far behind in terms of AI infrastructure. The largest computing power and technological platforms are in the United States and China. This makes European organisations dependent on technology over which they have limited control. 'If you look at patents, America has such a huge pile, China such a pile and we only have such a pile,' says Jonker, holding his right hand ever closer over the table, until it almost touches.

Trust as a foundation

Trust emerged as a key concept for successful implementation of AI. Organisations must be able to trust the provenance of data, the operation of algorithms and the reliability of models. 'Everything rests on trust,' says Gregory Verlinden, vice president Data & AI at IT provider Cegeka. 'Without trust, no adoption, no use.'

That trust is built through governance: clear rules, data classification and control mechanisms. Especially in critical applications, such as in healthcare, the financial sector or government, transparency is essential: if transparency is lacking, then trust will not come.

At the same time, there is growing awareness that data and AI models are increasingly intertwined, making the issue complex. In the words of IBM expert Ketelaar: 'To what extent can you trust a model trained elsewhere, with data you don't manage yourself?'

Economic value

Participants were unanimous on the need for investment in European and national AI infrastructure. Without such a strong, financial push, there is a risk of structural dependence on foreign technology and regulation. 'AI infrastructure is a basic need, just as electricity, water and railways are,' Verlinden said. 'If that falls away, then suddenly you can't fly anymore, you end up back in a walking pace. While the rest of the world is still flying.'

This calls for a combination of public and private cooperation. Examples from other European countries show that targeted investments in infrastructure, talent and ecosystems can lead to strong AI clusters. 'We should go much more for public-private partnerships,' Verlinden said.

For the Netherlands, according to the participants, the challenge lies in better direction and cooperation. While there is plenty of knowledge and talent, there is often a lack of scale and coordination to turn it into concrete economic value.

Scaling up to European level

Moreover, digital sovereignty is not even so much a national as a European task. Individual countries are too small to build competitive AI ecosystems on their own. However, sufficient scale and clout can be achieved through cooperation within Europe. 'We are actually just too small to set this up properly,' said Frank Ketelaars.

The AI factory being built in the province of Groningen was mentioned as a positive example. In addition to 60 million euros from the province, 70 million euros have now been made available for construction from the national government and 70 million euros from Brussels.

Photo: IBM