AI Development
Taking the Outside Lane

Europe has fallen behind in artificial intelligence (AI) development, but this disadvantage may soon become an advantage.

By Harald Willenbrock

What do European AI initiatives and legendary ski jumper Eddie the Eagle have in common? A surprising amount, as it turns out. Both started at the back of the pack in terms of resources, skill and experience. No one supposed that they might make any headway, as both faced opponents who outmatched them in almost every imaginable way. 
 
A quick look where the most important firms have been founded confirms Europe’s trailing position. In recent years, China has seen eleven AI unicorns – start-ups with more than a billion dollars in capital – and the USA is home to a whopping 28, but only four such start-ups are headquartered in Europe. Even Thierry Breton, EU Commissioner for Internal Market, was recently forced to admit that Europe seemed to have missed the first wave of AI business. Still, according to Praskant Khedekar, senior research analyst at Inkwood Research in Boston, the European AI branch comprising around 1,600 companies seems to be slowly catching up. He also noted that the European start-up ecosystem, with Great Britain’s around 500 Ai companies at its core, was clearly maturing as well. 
 
“Lack of financing remains a huge problem for the European AI branch,” Khedekar says. In 2018 and 2019, Europe invested around four billion US dollars in AI, while in China 25 billion US dollars and in the USA even 36 billion US dollars went into developing AI and machine learning over the same time period. 
 
This imbalance is reflected in research as well. Google far outpaces its competitors, ranking number one among the leading universities and departments for AI research and development in companies, followed by Stanford University and the Massachusetts Institute of Technology (MIT). The top 20 most relevant research institutions worldwide include China’s Tsinghua University (number 15), Switzerland’s Eidegenössischen Technischen Hochschule (ETH, number 17) and France’s Institut national de recherche en informatique et en automatique (INRIA, number 20) as the only three institutions not located in the USA. “China and the USA have developed enormous AI innovative capacity and are skilfully using AI to increase productivity,” experts from the Boston Consulting Group (BCG) say. “They are light years ahead of every other country in the world.”

A Lot of Gaps to Fill

So how might Europe bridge this ever-widening gap? A BCG expert team explored this question in 2020 and concluded that pretty much all the key elements essential to developing, testing, rolling out and commercializing AI technologies are missing in Europe. Some of the central issues include:

European firms lag behind in digitisation
In Germany the information and communication sector accounts for only around 1.7 percent of gross domestic product (GDP), while it makes up 2.1 percent in China and 3.3 percent in the USA. An economy that trails behind in digitisation is not ready for AI innovation either. 

The European market is fragmented
Although the European market for AI solutions is potentially large, European Union initiatives prefer to promote national solutions and regional hubs while the Chinese market successfully rolls out AI solutions ad hoc and on a large scale. The BCG analysts cite the digitization of the Chinese retail sector as an example, where online retail platforms Alibaba.com and JD.com have ensured that one-third of the country’s six million retail stores are now digitized and AI-supported in less than two years.

More system coordination
According to a 2018 survey by the European Commission, 60 percent of companies in Europe do not share data with others and almost as many do not use data they receive from others. This is partly due to strict European data protection regulations that focus on consumer protection and therefore restrict the distribution of personal data. So instead of a shared data ecosystem, lots of autonomous data islands have emerged, making deploying AI more difficult. It is therefore hardly surprising that none of the global tech giants such as Google, Amazon or Facebook are European. 

More data needed as raw material
“China has recognised that data is a critical factor in an AI-driven world and has encouraged the development of a data-labelling industry,” BCG experts write in their report. This involves categorizing and sorting data and creating large data sets. Just one of China’s most important data labellers, MBH, now employs 300,000 people involved in identifying and marking data to serve as raw material for AI algorithms, they say. 

Infrastructure? No so’s you’d notice.
After massive investments in infrastructure, China now has 40 percent of the world’s quantum computing capacity, which is far more powerful than classical computers in tasks like pattern recognition. This is three times more than in all European countries together. According to BCG analysts, even the EU’s latest Quantum Flagship initiative, which is releasing one billion euros for quantum computer development over ten years, is unlikely to change this much. By comparison, China is currently investing eight billion euros in its National Laboratory for Quantum Information Sciences.

Pivot Rather Than Race to Catch Up

In many respects, Europe seems to be taking a backseat to China and the USA when it comes to AI development and AI use in everyday life. Yet the reasons Europe trails behind, such as stricter data protection, are by no means all negative. Karen Yeung, Senior Fellow of Law, Ethics and Informatics at the University of Birmingham, says: “It’s up to the European public to discuss and decide what values we believe in, what practices we think are acceptable, and what kind of society we want to live in. And then we should live by our principles, our norms and standards.”

In other words, it is up to Europeans to decide which foreign AI applications they will allow inside their borders – and which they will develop themselves according to their ethical standards. This is precisely what the EU Commission is looking to promote with its AI strategy adopted in February 2020. 2.5 billion euros are to flow into launching trustworthy data platforms and AI applications as part of the Horizon Europe research programme designed to establish the continent as a “pioneer in trustworthy artificial intelligence.”

The hope is that Europe could take a “third path” in AI alongside state-controlled China and Big Tech fuelled US – one that prioritizes consumers’ interests and the protection of their data. “In the short run, the stronger European regulation compared to China and the US in this field might decrease our ability to scale revenue,” Camilla Rygaard-Hjalsted, chief executive of the nongovernmental organization Digital Hub Denmark, says. “However, in the long run, our focus on AI for people can serve as our competitive advantage and we can become a role model for the rest of the world.”
 
It remains to be seen whether this strategy will bear fruit. After all, in an industry where innovation runs on the essential fuel of freely available data, data protection is an inevitable competitive disadvantage. Europe is entering the technology race with a serious handicap, says Mark Scott, technology correspondent for the U.S. newspaper Politico. Nevertheless, the EU should not give up its values in the great digital competition. “Will Europeans be better protected online than others as a result? Yes. Will it help the region catch up to China and the US? In all likelihood, no.”
 
However, one of the advantages of being an outsider is that there is great potential for catching up. According to forecasts by the European Information Technology Observatory (EITO), total sales in the European AI market could rise from around three billion euros in 2019 to up to ten billion euros in 2022. The title of the EITO study is correspondingly hopeful. In keeping with the spirit of Eddie the Eagle, it is entitled: “Ready for Take-off”.