If EU states are to lead AI, they must not compete with each other

14 May 2018

Image: posteriori/Shutterstock

With the US and China dominating in the field of AI research, EU member states are starting to realise that competing among one another will only lead to failure.

While the biggest rivalry for the second half of the 20th century revolved around who had the biggest number of nuclear weapons, the rivalry at the beginning of this century is likely to be who has the most sophisticated artificial intelligence (AI).

So, there can be no doubt that two of the biggest players of this showdown are the two largest superpowers: the US and China. They are regularly talked about as having major ambitions to lead the technology within private companies and the public space, but also militarily.

The speed of development in China in particular has been fascinating to watch from the outside, with many observers describing its ascent to technological prowess as ‘China’s AI awakening’.

However, amongst all the bravado and chest-puffing, another player is attempting to make its mark on the game: the EU.

Unlike the US and China, of course, it is a coalition of states rather than a single entity, which on the surface would suggest it faces an insurmountable uphill battle.

A pan-European collaboration

However, this year alone we have seen the EU come together to flesh out some collaborative efforts, with this year’s Digital Day on 10 April seeing ministers from all of its member states hash out a plan on working together more on AI, particularly when it comes to establishing high standards of data protection.

At the time, Mariya Gabriel, commissioner for the digital economy and society in the EU, said it would “bring the digital cooperation in Europe to a new level regarding AI” and encouraged “all EU member states and stakeholders to contribute to our efforts to keep Europe a global player in the digital age”.

Away from the spotlight of politicians, a workshop held in January of this year saw academics and those from industry identify opportunities for pan-European collaboration in AI.

Among those attending on behalf of Ireland was Prof Barry O’Sullivan, founding director of the Insight Centre for Data Analytics at University College Cork and deputy president of the European Association for AI, and he recently spoke at it@Cork’s Tech Summit.

With 20 countries represented, the report authored by Charlotte Stix was able to put together an initial overview of the European AI landscape, covering the academic, industrial and funding ecosystems as well as governmental initiatives, with suggestions for the way forward.

Giving his overall thoughts on how it went, O’Sullivan is definitely in agreement that Europe is trying to establish its own identity when it comes to developing AI, especially with initiatives such as GDPR.

Drawing comparisons, he points to the US where Silicon Valley’s giants have gone hell for leather to develop AI capable of harvesting as much data as possible, arguably without thinking of consequences down the road.

Meanwhile, China is also attempting to find ways to harvest data en masse to meet a population of more than 1bn people, in order to cover everything from security to general internet use.

Fears of brain drain

“I don’t think a European would actually use those systems that integrate everything from your professional profile and personal lives,” O’Sullivan said.

“By and large, the consensus in Europe seems to be that Europe will develop AI technologies that have European principles and culture built in with respect for the citizen, as well as data protection, privacy, respect for lifestyle choice and other fundamental freedoms.”

But is the EU ecosystem lucrative enough to pursue that?

This has played into a fear among contributors to the report was that Europe is facing a ‘brain drain’ whereby the continent’s leading minds move to the US and China for more lucrative offers.

For example, a discussion among the workshop’s participants hosted by O’Sullivan raised the issue of bureaucracy in the EU, whereby an AI start-up takes six months to set up in France but only a few days in Silicon Valley.

Other findings included that the majority of countries mentioned distributed, but cooperative, AI clusters across Europe as a suitable next step to further the technology, such as the Benelux Association for AI and the Nordic AI Artificial Intelligence Institute.

And, while Europe has a very strong academic landscape concerning AI and AI-related research, the report said that there are vast differences in public acceptance, usage and uptake of ICT technologies across European countries.

AI for defence

One of the most controversial points surrounding AI is its use for the development of defence applications.

“One country specifically mentioned AI as a technology for defence applications,” the report said, “but there are significant concerns within the AI community regarding the development of lethal autonomous weapons systems, with a view that these should be banned from development.”

For O’Sullivan, it seems the topic will be debated for years to come.

“Politically, among representatives of member states, you don’t tend to find [the topic of military use] come up as it’s often regarded as something that’s controversial or something where there can be massive issues,” he said.

“For industry leaders and academics, the general argument is, we shouldn’t be developing these things.”

Based on what has come out of the workshop, Europe may need to hasten its pace of development or face not only being left behind, but stripped of its best talent.

Updated, 3.30pm, 18 May 2018: This article was amended to include the name of the author of the EU workshop report.

Colm Gorey was a senior journalist with Silicon Republic