EU Policymakers · ATLAS
Malgorzata NIKOWSKA
European Commission · Head of Unit · CNECT
Policy topics Malgorzata NIKOWSKA is active on
What Malgorzata NIKOWSKA has said (11)
- “Thank you very much and good afternoon, everyone. So it is my pleasure today to present to you our AI Contingent Action plan, and we have some slides which I hope will appear soon on the screen. Indeed. So let us go immediately into the first slide. So the the Commission communication on AI. Action plan was adopted on the 9th of April. And it focuses on five main elements. The first one is the infrastructure that we need to become a leading continent in artificial intelligence. The second element is data. That, of course, is the fuel for our AI models and system. The third element. Actually here presented as fourth, is the development of algorithms and, um, adoption. Uh, here we mean simply deploying the AI solutions across the various sectors. Then we present the element of skills that are needed both on the technical level of the artificial intelligence, as well as across sectors, to make sure that our companies can use the technology to their benefit.”
EU research funding
- “Thank you very much. Um, so I will try to go in order. Um, in terms of simplification, um, we are really looking first at the implementation of the AI act and trying to see how we can facilitate that through additional guidance and also through the AI act service desk, where any company will be able to ask their question and get an answer according to their needs. In addition to the AI act service desk, as I mentioned, we already have the network of European digital, uh, innovation hubs and they are also gearing up to provide this service, at the very least, to be able to tell the company whether they are at all concerned by the by the act provisions or not at all. And we estimate that 80% of the companies are not concerned at all because they are not in the high risk areas, and they can safely use the AI systems that they that they are using. The digital omnibus itself is planned more towards the end of the year, so there will be still further work. Part of it will be the result of our of the analysis of the ongoing public consultations. So please, by all means, encourage all the companies interested to participate and to make their voice heard as this will lead us to, to any potential, um, adjustment. But the whole we will be looking at the overall digital acquis for this digital omnibus in terms of procurement, it is indeed of key Importance. And that was already mentioned that in the competitiveness compass that that was adopted earlier this year, before the continent and in the continent, we actually repeating the language that especially for the public procurement, we will be looking to, uh, to, to to prioritize, uh, EU, uh, providers.”
"Buy European" provisions
- “And another element that was included under the infrastructure part was the presentation of the upcoming um cloud and AI Development Act. This act will will be coming towards the end of the year, and it will focus on ensuring that Europe is on par in terms of data centres, which can also provide computing solutions to our companies. Um, I should mention here one element, uh, namely, um, resource efficiency of these data centres and supercomputing infrastructure. Um, they of course are infrastructures that require quite a lot of resources. However, the green aspects are very much taken into consideration. And if you look, for example, into the call for expression of interest for the gigafactories, you see that the resource efficiency is at the forefront. And I'm happy to say that the European supercomputers are actually at the top of the green 500 list of top supercomputers in the world. So while these these infrastructures definitely need energy and resources, their efficiency is very much taken into consideration. Next slide please. The second element of the continent action plan was the focus on data. And it has um, presented the beginnings of the data union strategy, which will also be a communication presented by the Commission in the second half of this year, Yet, but already in the I action plan, you find the reference to, um, to the data labs, uh, which, uh, will work very closely with the AI factories and ensure that our companies have access to good quality of data.”
EU digital & tech sovereignty
- “And here we are looking at a number of guidance standards and of course code of practice which are in preparation to facilitate this Implementation. Um, we are going to set up and, and I act desk where, uh, companies will be able to get advice on the implementation of the AI act in their case. And we also have an AI packed community, uh, which I encourage everybody to, um, to get involved in because they can get all the information that they need of what is happening, the latest guidance that has been published, and any clarifications that we are providing in parallel in our public consultation, there is a number of questions relating to simplification where we ask companies to let us know what bottlenecks they see in terms of implementation. Next slide please. I should also mention that we are working to set up the regulatory sandboxes, which are specifically mentioned in the AI act, to facilitate the implementation. So um yes. Next slide. Thank you. So as I mentioned, we have currently three ongoing public consultations on apply AI until the 4th of June and also on cloud and AI Development Act and then opened slightly later in May. The data union strategy. And also there is open the call for expression of interest for AI Gigafactories. If anybody is interested to submit the proposal and discuss it with us. This is for the moment. Uh, initial discussion on how the AI factories can be set up, and I think that is all from me. Thank you very much. Thank you so much.”
Artificial Intelligence · EU Single Market harmonisation
- “If I can ask for the next slide, uh, data is, of course, um, a key element for, for AI. Without data, we cannot train our models and, uh, fine tune our systems. And therefore, uh, my colleagues working towards the data unions strategy are focusing on these so on providing access to good quality data, but also looking at other options for ensuring that, uh, European, uh, developers have access to these data. They are also looking at simplification of the digital rules. Overall, the digital acquis will be revised. Uh, well, our advice is the wrong word will be looked at with a view to see if we can make it better and simpler and more fit for purpose. And last but not least, they are also looking at data from the geopolitical context that, as we know nowadays, is quite challenging. Next slide please. They are very much now focusing on collecting the right information for the data union strategy. So they are carrying out public consultations. And we very much encourage you to take part yourselves, but also spread the word about those consultations they are carrying out. Sectoral workshops and also organizing a number of outreach events. It is very likely that the Communication on Data Union will be coming out together with the apply strategy that I will be telling you more about as a next point. Next slide please. So the third element of our I continent action plan is to apply an AI strategy that will come more or less in autumn this year and build on what we already have.”
EU digital & tech sovereignty
- “And last but not least, we are looking at regulatory simplification and I will walk you through with a little bit more details of these five elements. So can I please ask for the next slide. So you probably have heard about the AI factories. Uh, these are the compute infrastructures that, uh, already base themselves on our existing HPC network, which is the best world network of public supercomputers. And the AI factories will be not only providing the compute services, but also access to data and other services that our companies in particular start ups and scale ups might need uh and uh. In accompanying the AI action plan was also an update of the access condition to these infrastructures to facilitate and speed up access for our companies. Uh, the second element, um, building on that, uh, you heard about this as well is the AI gigafactories. Uh, these we don't need, uh, in every country, but we need them to be available for European companies and European researchers, uh, to provide access to really top notch models. Uh, if we compare currently, our supercomputers already available have more or less 15,000 of these AI processors. The AI factories will be up to 25,000, while the Gigafactories we expect them to reach 100,000 of these AI processors Accompanying the AI Contingent Action Plan was a call for expression of interest for the potential consortia that are interested in building these gigafactories in Europe, and we look here very much at funding from the private sector, so not only from the public, but also from the private sector.”
Artificial Intelligence
- “And that will also be, uh, be looked at in terms of I and already now we have one of the JNI calls for public service. And it looks at EU solutions or solutions, uh, developed uh, in uh, in Europe. Um, in terms of uh data union, uh, strategy. So, um, the main objective of this strategy is indeed to ensure the availability of data for the AI I a training of models and then for the deployment of the systems and colleagues here. We are currently on carrying out a number of uh, well, first of all, they are carrying out the public consultation, but also have a number of actions that I mentioned to to gather this input also across the various sectors. And we are doing a similar work on on applied strategy to see how to encourage the uptake. And the issue of data is often mentioned, but surprisingly, maybe not that often in terms of complexity of the regulation around it, but rather of simple access and the difficulties of interoperability, for example, among different, um, producers of machinery, that then you have two types of a machine, of two types of machines from which you would like to have the data, but they are produced by different producers, and therefore the data is in different formats.”
Artificial Intelligence · Interoperability requirements for digital platforms
- “And already that is a difficulty to then be used for the for the I development or deployment. Energy efficiency is of a key concern. Even if it was not our priority, I think we all see what is the situation now. There is simply there are simply objective difficulties, both in terms of energy per se, but also other resources like for example, water. And a lot of attention is being paid to to these. A number of our supercomputing centers already today have a closed circuit cooling or are testing solutions that allow for this. They are looking at the most energy efficient chips to be deployed so that they use less energy. But of course these are still large infrastructures. So there is a certain number of, um, certain energy needs that will have to be met. And also in the context of the cloud and AI act, this will be looked at to ensure that these uh, doesn't harm and uh, is all the decisions are taken in considering the capacities available. Um, as I said, we are looking very much at supporting EU or European solutions, be it for the deployment or the development of AI. And here again, we are having, on the one hand, the network of European digital innovation hubs that already today are supporting the companies developing and deploying AI so that, you know, there is also knowledge about the availabilities of AI produced in Europe, because what often happens is that companies are simply not aware of the solutions that are already now available.”
Energy efficiency
- “And here we are also looking at public sector to promote these and to be aware of these solutions. And in addition to the the support that you can already today get through the network of European Digital Innovation hubs, we are looking also at. We have created a on demand platform where these solutions can be in a way advertised, which will be a marketplace for solutions made in Europe that companies, but also public sector organizations can can see first and use, test and potentially procure for their needs AI factories. If you have ever been to a server room, then a supercomputing center. It's like a huge server room. Lots of lots of these kind of big towers with blinking lights and cables. And they already exist, uh, across Europe in many different places. Many of them are in the top, um, best supercomputing centers in the world. But not all of them are optimized for AI use. So what happens now with the AI factories is, on the one hand, we are looking to upgrade these infrastructure Structure to be optimized for. I use it's a different type of chips that are better suited for AI needs. So this is one element. But this is not all what the AI factories are about. The AI factories are also to create an ecosystem around this. So not only to have the machine doing the compute, but also have access to data support for the companies to develop and facilitate the access to use the computing time to the to the maximum effect without wasting time for example, for for testing.”
Artificial Intelligence
- “And already today, the company can really go to the Adi and get the support that they need. They can ask the question about the the AI act if the app is not able to help them. Sorry. Then they can guide them to another hub maybe, or to another, um, to another support, uh, tool, like for example, the AI factory. But also they can come to us and ask a question to the AI service desk and get the answer that, uh, that will be needed. And this support is available at a local level, which is very important because many small companies will have difficulties, for example, even on linguistic level. And they're the hubs speak the language, know the local condition, and they can really advise them best on on how to get the support they need. And that support is available already. Uh, today. And last point on how do we encourage uptake. It is indeed, um, something that we are thinking very much about and what we are looking at. So we are actually trying to have this approach based on use cases. So we are looking at what type of application works best in a sector which has a relatively acceptable and feasible requirements, will have the highest impact and can be scaled up or deployed most widely. And those application areas will be promoted in our applied AI strategy, because we really want to see the European companies making a huge leap in terms of AI deployment. Thank you.”
Artificial Intelligence · EU Single Market harmonisation
- “So the notion of AI factories, the Gigafactories, but also our network of European digital innovation hubs, which are one stop shops for any company to get support with their digital transformation. They are present in all member states as well as a number of associated countries, and they already today provide the services that the companies need in terms of digitalization in the future. And we are currently evaluating the second call. They will also become AI experience centres, and all of them will be providing services relating to AI according to the needs of the company. So some companies will require very sophisticated services, other maybe some more basic services, but all of them, all companies will be able to get the support they need. So based on what we already have, we are looking on how we can increase the uptake of AI across the sectors. As you know, currently the numbers are not too great, around 13% as per latest Eurostat figures. So we are really looking to increase this across the key sectors and the number of sectors. 13 were mentioned already in the AI Continent Action plan, and we are also having public consultation open right now. And possibly this number can slightly be um, be be changed. We will also be looking how to increase the uptake of AI for the public sector, which as we know, not only, um, should deploy the technology, use the technology more frequently, but it's also an excellent source of spreading the information and making sure that that companies, but also citizens have more trust towards the technology.”
Artificial Intelligence · Digitalization of public governance & administration