“The problem for AI in Europe is not the money, it is finding talent”

Holger Hoos, co-initiator of CLAIRE and Professor of Machine Learning at Universiteit Leiden (07 September 2018, CLAIRE Symposium, Brussels)

The CLAIRE symposium is a rare moment of bottom-up self organization: Agile, urgent, and with a sense of mission. On 07 September, 100 key community members from academia and industry came together to discuss and agree an action plan for all of AI, all of Europe, and with a human-centered focus.

It was my pleasure to participate. This is the summary of my most important insights and results from the symposium. I have identified the following key areas:

  1. Talent search and retention
  2. The relationship of academia and industry
  3. European competitiveness

“And then she got an offer in New York, and left London.

CLAIRE symposium participant on the better salaries, more interesting use cases, and more advanced practices in the United States

For many symposium participants (the younger ones anyway) the attraction and retention of talent in Europe was the key issue. AI is a human endeavor, and it is only as good, relevant, and useful as AI practitioners make it. Moreover, it is a new and interdisciplinary endeavour, so much of the talent has a variety of numerate backgrounds.

How to attract AI talent to Europe?

  • Be more visible at the global level by offering online resources and educational materials ‘made in Europe’ or ‘made by Europeans’.
  • Create much larger education and training programs to attract talent worldwide, stop limiting enrolment from outside the EU, and invite successful graduates to refer and recruit more talent globally.
  • Promulgate human-centered AI with its focus on transparency, accountability, and responsibility, and the interesting opportunity of creating more user-centered AI that empowers people.

How to train talent?

  • Create new and leading in-person AI courses from Bachelor to PhD that deliver insight and training at a level that is not available in online courses.
  • Have industry leaders show their most relevant use cases, ask industry experts to co-teach, and use industry funding (e.g. via a foundation) to scale high-quality training more rapidly.
  • Create large talent hubs outside rigid academic structures so that many talents can work on the same or similar issues and also fund them to be mobile for working and learning wherever in Europe suits them best.

How to keep AI talent in Europe?

  • Easy access to large-scale computing resources at all times, i.e. computational power equivalent to that of a researcher at a leading AI firm in Silicon Valley.
  • Enabling and encouraging dual affiliation in academia and industry not only for more competitive salaries and exchange of practice but also to ensure that the training of the next generation scales (e.g. France has implemented such a scheme already).
  • Offer better and more funding opportunities for collaborative projects at the Master, PhD, and Postdoc level to facilitate the emergence of better interdisciplinary teams and practices.

“Use cases provided by industry and startups often translate into the more interesting and challenging research questions.”

CLAIRE Symposium participant on why as a professor he spends a good deal of time interacting with companies

While the difference between basic and applied research matters to AI, industrial research is a major trend, and many AI firms have large inhouse research teams. Moreover, the opportuntiies for applying AI are growing rapidly.

A human-centered focus implies equally a user-centered focus, which, in turn, is what startups typically need for traction, i.e. to launch and iterate their product. Hence, if European AI scales it creates investment opportunities, spin offs, and new products.

How could industry become more engaged in funding or inspiring research?

  • Regular exchange at the working level between academia and industry increases the level of trust and collaboration, e.g. by sharing results, student placements, and working towards a joint ecosystem.
  • Engaging industry sectors, companies, and leaders most invested in AI and potentially benefitting the most from further advances will help develop the ecosystem and launch joint facilities.
  • Developing an actionable plan for large-scale funding of several European centers of excellence, each having at least one foundational and one highly relevant application area that is pursued in cooperation with industry.

Which industry sectors are most relevant in developing applied research?

  • Special about Europe is its world-leading manufacturing sector with opportunities for the further development of robotics.
  • France has chosen to focus its national AI strategy on health, the environment, transportation & mobility, and defense & security (report by François Sillion, coordinator of the French AI research program).
  • The European Commission has identified as performant sectors transport, health, and manufacturing (report by Juha Heikkilä, AI & Robotics, European Commission).

“AI is a global ‘game changer’ that has become a major driver of innovation, future growth, and competitiveness.”

from the CLAIRE vision for AI

I have met automotive experts believing that the autonomous driving developed in the United States and China potentially has consequences for the European automotive industry on a scale comparable to impact the iPhone and Android phones had on Nokia.

Andrew Ng has called AI ‘the new electricity’, and expectations are that AI will have wide ranging impacts on society and the environment. So what must Europe do to create the future? What are the next steps?

  • Infrastructure: Start seeding European facilities for AI with the aim to roll out computing power and operating capacity on a scale comparable to the US and China.
  • Data: Enable the creation and sharing of suitable data at scale, e.g. open government data, anonymized medical data, pooled automotive data.
  • Hub: One or more significant places for co-working and the co-creation AI at scale — underpinned by a world-class infrastructure and accessible data.

The above represents my personal summary of the CLAIRE Symposium. My special thanks go to the organizers Holger Hoos, Morten Irgens, Philipp Slusallek, Silke Balzert-Walter, and Alexa Kodde.

You can sign and share the CLAIRE Vision here.