TechTalk ep17: On Luxembourg’s national AI strategy

Paramita: Hello and welcome to PwC Luxembourg TechTalk. Luxembourg has recently come up with a national artificial intelligence strategy to provide a framework for AI to work in. Today, I speak with Laurent Probst, partner and Andreas Braun, senior manager about this national framework and how such a strategy helps create the AI value chain.

Paramita: Hello we have in our studio today Laurent Probst and Andreas Braun. Good to see you back Andreas. And they're here today to talk to us about the artificial intelligence strategy that Luxembourg came up with on the 24th of May. But first why do we need in the strategy Laurent?

Laurent: Thank you.

I think today artificial intelligence technologies are too important to be ignored and I see five reasons why Luxembourg needs such a strategy.

First of all, AI technology will impact the life of people as a citizen when we will use health services in the hospital or when you will ask for a new authorisations, new business licences for instance or as a professional when you will work into a company that will integrate new AI technologies. So our life will be impacted, all our lives will be impacted. That's the first reason.

The second reason is that it will modify our social fabric meaning that tomorrow rather than contacting someone on the Help Desk or talking to someone we'll talk to a robot.

What does it mean? How that will change our behaviour?

We have to think about it and to look at it. The third reason is when we look at all these things clearly it will have an impact on trust. Do we trust an algorithm? Do we trust an AI tool or not? And it is very important to keep and to increase the level of trust in our society today. So I think that I would say that's why building trust is the key role of this AI strategy. As a fourth reason, I see the need for assisting private companies, individuals and also administrations to adopt in a safe way, in an efficient way these technologies. And clearly to be able to upskill their people, to facilitate the adoption and the understanding of the potential benefits as well as the risk of such solutions.

And last reason is about the competitiveness of the country.

Today it's very important to develop an AI strategy to figure out the new set of investments on the private and public side and to position the country as, I would say, as a testbed for AI and to be able to attract very important investment which would be made into these new technologies. So overall I think that these technologies are so important for human beings that it's very important that a country takes it at the national level and embrace all the different dimensions of such a technology.

Paramita: So would you say that it's actually giving artificial intelligence a proper framework to work in? That's where Andreas what do you think that is where countries are heading to when they think of having an AI strategy?

Andreas: That is exactly what most countries are working on right now. It's about how to deal with governance aspect, how to deal with the innovation aspects coming in all. And of course also deal with the societal impacts as Laurent already mentioned. So it's about creating this overarching framework and always with the various country specific components to foster the strengths of specific countries. So for example, in countries like China obviously a lot of basic research. In countries like Germany, we see a lot of leverage in Industry 4.0 topics. Whereas in Luxembourg of course we have quite a few topics that go into the financial domain and how to leverage the innovation there. So it's really also about creating this framework but making it country specific so we can best make use of that in the future.

Paramita: And since you're talking about country specific what does the Lux AI strategy consist of?

Laurent: The Lux AI strategy is firstly focussed on being a human centric strategy.

I think this is key. And we have to develop I would say the framework which will enable to really develop a very beneficial strategy for all.

This is a key important dimension. This AI strategy is I would say an international strategy.

It takes into account the regional dimension of Luxembourg leveraging also all the regional assets as we have around the country a lot of interesting research or AI companies that's very important to take it into account. And also leveraging a lot of international partnerships with countries or with private companies. So by taking immediately an international dimension this strategy will really leverage the best.

I think that's a very important distinctive aspect of this plan. The objective is also clearly to leverage the investments and the project, which would be made on the private as well as on the public side. And together it will clearly improve the competitiveness of the country. Today, Luxembourg is not in the best position in terms of e-government and this strategy will clearly foster the positioning of Luxembourg and I would say the outcomes of Luxembourg in this field and that will help administration to transform, to adapt and to face the increasing demand of high quality public services from I would say any citizen. I think one of the very interesting points of the Luxembourg AI strategy is to take Luxembourg as a testbed.

Paramita: What do you mean?

Laurent: Meaning Luxembourg could be considered as really a site where we could operationalise all I would say the key concepts of AI meaning design by security or security by design, ethics by design from the development of an AI application to the provision of the service to the end user.

And this is very important to operationalise this, providing all the assurance of a good service checked by the government, checked by the regulator, monitored all along the chain. Luxembourg could be one of the first countries to provide such an infrastructure, such a set of services to be able to have a complete transparent overview of the functioning of AI applications.

Paramita: Right. The strategy, does it have any aspect of regulation, any aspect of ethical AI... does it speak of that?

Andreas: The strategy definitely also has that in mind. So in terms of regulation, of course it's important to set up the infrastructure, the discussion frameworks on how can we regulate AI in the future also in Luxembourg. And here we are also linking to European initiatives in there. So the European Commission has already released ethical guidelines for the application of AI and it's not really about reinventing the wheel it's about how can you take these works that already have been done and apply to your country specifically.

So experts will be involved in this process from different domains not only technical but of course also social and people from the administration and they will see on how can we take these existing guidelines and apply this country specific framework from there.

Paramita: OK. And Laurent just to understand because when you spoke about the testbed thing where does the strategy come in? I mean the government provides the infrastructure is that so?

Laurent: It's more than an infrastructure. It's a framework. And when we are talking about regulation we have to think also about innovation.

So we have to be careful about this new regulation in order not to kill innovation.

Providing a testbed will be a key facility, a set of methodology, a set of competencies, a framework which will enable private and public stakeholders to understand the real benefits and the real risk of different applications. Without this real case, it's very difficult to understand the benefits and the risk of new applications. And this is about also developing our common understanding and that will provide a very strong competitiveness advantage for Luxembourg knowing that we will develop this common knowledge.

And when a private investor, private company will come it will immediately recognise the knowledge of the different stakeholders about this topic. And so that's a very important asset for the future of the country. It's not easy. It's a challenge, it's very complex. But I think that we see from the discussions with different stakeholders here... with the university, with research centres, some companies, the potential benefits of developing such a testbed for AI. And this testbed could apply to different domains.

We have already a first case with autonomous driving. We might have another one on financial services or on health applications. So we have to decide in which domain we want to develop some verticals.

But I think that that's a very important case.

Paramita: And my next question, which is linked to this, is how do businesses... how are businesses impacted are enabled by this strategy in Luxembourg?

Andreas: So the strategy followed a typical process in its creation. There were various stakeholders consulted in the creation of this strategy.

Various industry leaders that have a strong link to AI have already been interviewed and their inspiration and feedback have been put into this document. And as a next step, the government plans to set up various working groups that will facilitate this translation of an AI innovation and AI regulation into the typical business culture within Luxembourg.

So we foresee that in particular in the financial sector when AI applications are becoming more commonplace that regulators such as the CSSF will support these kind of processes and that also industry experts will continue to contribute to developing the strategy even further.

So it's maybe an interesting topic that also the AI strategy in the next step will be performing a public consultation. So all residents of Luxembourg or even interested people from abroad are invited to give their feedback on the current version of the document, which will later be translated further into more elaborate a document that will be likely coming out in the fall.

Paramita: That sounds really interesting.

Laurent: Also from a business perspective, I think today the two main issues for businesses are first of all a real in-depth understanding of the quality of data they have before adopting the right AI strategy.

This is a key milestone for them and most of the companies are not yet there.

And the second point is about understanding which are the real benefits and the risks of AI for their own activities. Knowing that for most of them they already are buying tools which include some AI components. And when tomorrow you buy a banking software, this banking software for a bank has already some AI components. For them it's very important now to get a real understanding of what they buy, how do they buy it and to get the right data and AI governance at the cooperate level. And linked to that when they will develop further applications with data what we will need is they will need access to more computing power and this is very interesting to link that to the project of the Luxembourg government of a high performing computer which will be I would say installed in Luxembourg in 2020. And that will be a new major digital infrastructure available for companies and that's very important for us, for our future to get this kind of computing power.

Paramita: Which other countries have a strategy like this?

Andreas: So AI strategies are kind of popping up all around us these days. There are some front running countries I would say. So in Europe most notable are Germany and France that already in 2017 and 2018 have worked on their strategies that include also significant investments by the government into research and innovation. In Asia, we had China and Korea as countries that were going into this domain very early. Very targeted strategy in the case of China, very innovation focussed in the case of Korea. And if we look across the Atlantic Ocean, Canada already also has a country strategy for artificial intelligence. The United States even though they are one of the AI powerhouses so to say they don't really have a national strategy. So there it's a more diverse topic. So the Department of Defence has its own AI strategy. Then the research-oriented ministries have their own strategy. So it's kind of not very country focussed but still they are in the lead just because their companies have access to very good data and they have access to a very large group of talented computer scientists and engineers that are working on those topics. And to complete our nearest neighbours Belgium has also released its strategies just a few weeks ago and all European Union member states are supposed to create their own strategies within this year. So we will see more strategies coming up. And ideally they are all linked within Europe at least to the strategy of the European Commission overall. And Luxembourg did a pretty good job of also linking the country's strategy to the European overall goals because as a single country it's really hard to compete against powers such as the United States or China. But given the kind of European landscape if everybody works together and optimises their investments there is a much better chance for Europe as a continent.

Laurent: And I think trust will be the key and the key success factor of national AI strategy will be in the way a country is able to operationalise this strategy and build a trusted environment and that will be key. It's operationalisation and trust. It's not just about having a nice paper with fantastic I would say key words but it's how to link the stakeholders together, how to build up this artificial intelligence value chain for the benefit of all. For all.

I think that that's the key. That's really the key point out of it because AI is about a new social fabric. It's not about just the economics. It's about a new social fabric. And if a strategy does not include that I think we will miss the point.

Paramita: Yeah. And I think that trust was at the heart of all these strategies coming up in the first place and it will be also like you said the and objective of these strategies. So on that note I think we will end our conversation. So thank you so much.

Laurent and Andreas: Thank you. Thank you.

Paramita: So that was Laurent and Andreas discussing the Luxembourg AI strategy. Tune in next week for another brand new episode of TechTalk.