Europe ‘losing ground’ to US and China on AI research

Europe faces a serious threat to its future prosperity and the safety of its citizens if it does not act soon to properly coordinate its publicly funded research into artificial intelligence, a leading professor has warned.

Holger Hoos, professor of machine learning at Leiden University, told Times Higher Education that the continent was “losing ground every week” to the US and China in the discipline, describing the lack of decisive action at a governmental level as “disturbing”.

Professor Hoos, who works at the Leiden Institute of Advanced Computer Science, is one of the driving forces behind the Confederation of Laboratories for Artificial Intelligence in Europe (CLAIRE), which is pushing for formal pan-European cooperation on AI research, including establishing a central hub that would have a global impact similar to that of the Cern lab for physics in Switzerland.

He believes that without firm action, European universities will continue to lose top AI scholars and students to industry and institutions in other parts of the world, particularly in the US.

But Professor Hoos was also concerned that the current fragmented nature of public research on AI in Europe risked ceding too much control of the discipline’s ethical direction to large multinational companies.

“We have been letting, to some degree, industry run with it for too long already. The public has an important stake in this, and therefore the public should invest – if only to ensure that we have free access to the same kind of cutting-edge AI research,” he said.

Professor Hoos added that there was a risk of Europe becoming “dependent on AI technology developed elsewhere”, which “could be a very serious financial and economic hindrance to the prosperity of Europe”.

“If we were completely dependent on aircraft or automobiles or mobile phones produced elsewhere, it would not be advantageous to our economy. With AI being such a fundamental set of technologies, it is even more important that we do have a stake in this.”

As well as a Cern-type central hub, the CLAIRE “vision” is also to massively ramp up funding for European AI research, strengthen existing research networks and identify major regional labs as “centres of excellence”.

It has already gained support from more than 1,700 academics across the continent, from national governments including Italy and the Czech Republic and from some 700 AI experts working in industry.

While he recognised that reaching such “ambitious” goals would be “unimaginable” without European Union backing, Professor Hoos argued that, as is the case with Cern, cross-national cooperation “should be not be limited to the EU and the [European] Commission”, especially given that Brexit would leave the UK – a major player in AI research – outside the bloc.

He also said that although the commission was thinking about the challenges posed by AI “very carefully” – last month it published “ethics guidelines for trustworthy AI” drawn up by its High-Level Expert Group on AI – “honestly, I am hearing too little from them, and what I am hearing is by far not ambitious enough”, especially given the ongoing talks over the next framework programme for research to replace Horizon 2020.

If action was not taken, he said, the drain of talent would worsen and Europe risked not having the expert capabilities to deal with the ethical and safety problems posed by AI.

“Considering the rate at which [Europe is] bleeding talent and our decreasing share of global scientific publications in the field…I firmly believe we are losing ground in [the area of] AI every week. It is disturbing not to see more decisive action,” he said.

“My main worry about AI is that because of the shortage of talent and the enormous demand, once we dip too deeply into the talent pool and have people who are too weakly qualified or unqualified to develop, deploy and work with AI systems that they don’t understand, then some pretty bad things can, and probably will, happen.”

Professor Hoos’ views were echoed by Thomas Metzinger, professor of theoretical philosophy at the Johannes Gutenberg University of Mainz, who represented the European University Association on the European Commission’s expert AI group. Although he believes that Europe has taken the “intellectual lead” on AI ethics, he has criticised the guidelines as being too industry-dominated.

Professor Metzinger said it was vital that top humanities academics such as the next generation of philosophers coming through were able to work alongside computer scientists if major research hubs were set up.

“Ideally, young philosophers should be co-located [with the computer scientists] so they see what real cutting-edge research is,” he said, adding that “governments in Europe literally need hundreds of really well-trained experts who know…what is really going on in AI.”

simon.baker@timeshighereducation.com

Academics and university leaders will discuss how universities can encourage more consideration of artificial intelligence in teaching and learning practices at Times Higher Education’s Teaching Excellence Summit, which is taking place at Western University, in London, Ontario, Canada, from 4-6 June 2019.