Eighteen months ago, over 2,000 European researchers called for major and coordinated investment into AI, where the focus would not be on technology but on humans. The call was spread through the newly formed European network CLAIRE – The Confederation of Laboratories for Artificial Intelligence Research in Europe.
Ursula von der Leyen, president of the European Commission, presented in February a strategy for AI priorities within the EU. It contains not only increased co-ordination of research but also support for AI to contribute to solving challenges faced by both individuals and society, fully in line with the call from CLAIRE.
“An important part of CLAIRE’s vision is also to create an AI hub in Europe. This would be the focus of a critical mass that offers the best research facilities – a CERN for AI. This is also part of the AI strategy laid down by the commission, as a ‘lighthouse’”, says Fredrik Heintz, AI researcher at LiU. He is member of the European Commission expert group in AI, and director for WASP graduate school.
Human centered AI
He is particularly happy that the EU agenda strongly advocates for AI that puts humans at the centre.
“Fundamental human and ethical questions are important now. The work to study how AI influences the humanities and society, such as the WASP HS (Humanities and Society) Fredrik Heintz Photo credit Anna Nilsenproject here in Sweden, does not have any equivalent in other European countries”, he says.
Establishing an EU strategy is a first step, but European researchers are still anxiously holding their collective breath. The results of a major call within Horizon 2020 for setting up networks of centres of research excellence throughout Europe, four networks each with a particular focus, are to be announced within a month or two. Fredrik Heintz is coordinator for one of them, TAILOR, which stands for “Foundations of Trustworthy AI – integrating learning, optimization and reasoning.” A group of 55 participants in 20 countries has jointly submitted the application.
“We want to build a foundation for broad and reliable AI: we want to integrate and combine several technologies within AI, using both machine learning and reasoning techniques. At the moment, researchers are debating actively whether machine learning is sufficient, and we believe that it is not. This view is also shared with Ursula von der Leyen and the European Commission as they also explicitly state how important it is to combine learning and reasoning”, he says.
“We have a good chance of being one of the four selected, but we’ll have to wait and see. CLAIRE has also applied to lead the coordination and support action the four networks.”
Huge investments in basic research in Sweden
Fredrik Heintz is in many respects reasonably satisfied with developments so far; a great deal has happened in recent years, even though a lot remains to be done.
“Here in Sweden we have the huge investment in basic research under the Wallenberg AI Autonomous System and Software Program, WASP, and further investment in WASP Humanities and Society, WASP HS. It is truly great, also in an international perspective. But education and professional development are also important – to achieve lifelong learning – and there we still have a long way to go.”
The AI Competence project started two years ago, an initiative in continued and professional education during 2018 and 2019. The project has been coordinated by Chalmers University of Technology, and LiU has contributed several courses of commissioned education.
“We have had the opportunity to develop courses that we could offer as commissioned education. We gave three courses in 2019, and more are planned. There’s a clear demand for the activities developed in this iniative, but unfortunately it is unclear whether there will be a continuation.”
The LiU distance course “Elements of AI” has been extremely successful, originally developed by Helsinki University, with nearly 4,000 students admitted in the autumn of 2019. More than 900 of these have passed the examination that completes the course.
“A very good initiative, that has both increased knowledge about AI and raised the profile of LiU”, says Fredrik Heintz.
He is also happy about the AI courses that are given as part of undergraduate education at LiU and the way in which they are continuously being improved and extended.
Politicians uninterested
“AI of Sweden” is another project, with one node in Gothenburg and a newly established node in Stockholm. Linköping is also getting involved and the work is led from the Mjärdevi Science Park. The focus here is on increasing the application of AI in companies.
“Many projects have been started, but most have modest funding and a short-term perspective, so there’s a lot of overhead and the projects never really get the chance to take off.”
An AI agenda for Sweden is also being developed in a participatory process, where RISE is coordinating the process of formulating concrete goals and a way forward for AI in Sweden. Further, the Agency for Digital Government (DIGG) has recently published a report that showed that the public sector in Sweden alone can reduce costs by SEK 140 million through the use of AI.
“But this report hasn’t received much attention so far, and we find that politicians are rather uninterested in the topic. There’s no real political conflict here, and this may be why so little is happening in political circles. It seems that they regard it as essentially just another special-group interest when we try to discuss AI and its significance for social development”, says Fredrik Heintz.
He does, however add:
“We have made quite a bit of progress in Sweden, but in Brussels it’s often the Finnish initiatives that are discussed and praised. I think we’ve achieved more in Sweden, but we just don’t blow our own trumpet enough.”
Footnote: CLAIRE now has 3,343 member organisations and encompasses more than 20,000 people in 34 countries. The organisation’s full response to the AI agenda published by the European Commission is available at: https://claire-ai.org/wp-content/uploads/2020/02/CLAIRE-Press-Release-11.pdf
Translated by George Farrants
