The well-known big hitters in artificial intelligence are the United States and China with large investments and a lot of exposure about those investments. Furthermore, Canada, Australia and Singapore are also making headway with AI technologies.
However, Professor Holger Hoos, professor of machine learning at Leiden University in the Netherlands said that Europe is at the forefront of AI research output according to one of the largest computer science publishers.
He also said that a fair about of AI innovation taking place in Europe is being translated into industry, resulting in new products and processes. In addition, Europe-based laboratories have some significant competitive advantages.
"The talent pool is a big asset that Europe should activate."
The advantages can be exploited by the Confederation of Laboratories for Artificial Intelligence Research in Europe (CLAIRE), of which he is a co-initiator. One of those is diversity, both of culture and of values which result in robust solutions that can be successful in different geographies. Hoos said: “If you find a solution to a problem that works for a major part of Europe or even all of Europe, chances are very good that it will also work for a bunch of other societies and nations.” The second is talent pool particularly in Eastern and Southern Europe. “I see that as a big asset that Europe can and should activate.”
CLAIRE is a group of 341 institutes across 34 countries, founded in June 2018 which has recently relocated its headquarters to The Hague, in the Netherlands. It was set up 18 months ago to support AI existing experts and make sure that they can continue to excel in what they do and remain in Europe while doing so, which in turn develops the talent pool and strengthens AI talent in Europe.
"If we enable collaboration ... we can get a lot of additional value."
Another aspect of CLAIRE’s mission is to “exploit the synergies within this network,” said Hoos. “If we enable collaboration, particularly beyond boundaries, to a larger extent than in the past, we can get a lot of additional value and we are also in a much better position to develop global leadership, particularly regarding the use of AI to tackle pressing global problems, such as climate change.”
If the regional AI network is to encourage cross-border collaboration, how would this work if the two countries in which the labs were situated in had different attitudes to transparency within an AI algorithm?
Hoos said that it is important to recognise that there is a trade-off between time to market and being robust and unbiased. It is also important for the labs in those different countries to discuss those trade-offs. Germans, for example, are notoriously cautious about how they allow large corporations to handle their data. It would, therefore, be safe to assume that they would want any algorithms that make decisions based on that type of data to be as transparent as possible.
Staying with this examples, in Hoos’ view, the German AI researchers would be pressed to demonstrate that the particular algorithm they are working on can have a high level of transparency without any loss of performance. He said: “The real value comes having this discussion in the first place, exploring the trade-offs together and having parties that have different interests push each other to make sure that what is perceived as a trade-off really is a trade-off.”
I wondered how Hoos viewed the risks associated with artificial intelligence and how the CLAIRE network could help to mitigate them. He said that what he views as the biggest risk posed by artificial intelligence is not the same as what broader society perceives it to be. The former being weak AI failing us in subtle with the latter is strong AI taking over and replacing the human workforce.
Hoos said that there are ways that CLAIRE can help to alleviate these negative effects. The first is to make sure there are sufficient AI experts who can easily recognise when systems start to make problematic mistakes and act as a human line of protection against weak AI systems.
The second is to encourage research in the direction of creating AI systems that have fewer weaknesses and of creating AI systems that can monitor themselves and signal to a human decision-maker when an operation becomes problematic.
The third is to have judicious regulation. Hoos said: “Judicious regulation of AI and preventing the worse kind of mistakes one can make would be a useful thing and I am delighted to hear that the new European Commission seems to see this as a priority as well and CLAIRE pushes for this in a way that makes sure that we are not cutting ourselves off from promising directions of research and innovation.”
He added that although GDPR could be seen a controversial, it is beneficial and even China is starting to mimic parts of it because they can see the value in it.
With the UK currently and in the foreseeable future stuck in the quagmire of negotiations relating to a withdrawal from the European Union, what could this mean for cross border collaboration?
Hoos gave the caveat that he is no legal expert but this would be problematic for both AI researchers in the UK and their counterparts in other areas of Europe. He stated that the UK is strong in AI research by many measures and those conducting that research would be hindered in participating in massive research projects without aligning themselves a lot more closely to the US and China. He said: “It is also worrying to the rest of Europe because the rest of us would lose a very important player in this, not just in terms of data but also in terms of algorithms, talent, industry that takes this up.”
However, CLAIRE was not set purely as an EU endeavour. Hoos pointed out that his two co-initiators are based in Germany and Norway, with CLAIRE viewing Europe as a geographical region in the same way as organisations such as the European Space Agency.