We are pleased to show you the 9th edition of the European AI Forum. This is our flagship conference, which is produced in cooperation with the nine national AI associations in the EU.
yeah yeah hello everyone my name is Danel Lau I’m the president of the European AI forum and I’m very happy to welcome you to the ninth edition of the European AI Forum live from willus in Lithuania with me is my colleague cl cl is from H France are uh the French AI Association and our host Lenas from the artificial intelligence Association of Lithuania we are very happy that you’re here live in villus but of course the thousands of people who are watching us on YouTube and streaming and we’re very happy U to be here that was an very interesting year for artificial intelligence in Europe um only the highlights of course the the European AI Act and the national implementation which are which is really important also in the following years but when you he the first time uh at the European AI Forum we are in association of associations to speak we are eight countries working together to promote artificial intelligence in Europe and of course I am also um the general manager of the German AI Association we are representing 470 startups and scale-ups from Germany with um a focus on the development M of AI tools and so we’re working together throughout the national associations and perhaps CL perhaps you can um explain a little to us about for example H FR yes um hello everyone and thank you Daniel for the introduction um so what we do at up AI is that we Federate the French AI ecosystem we do it through working groups through projects through events and we now have more than 200 members yeah that’s not too bad Lenas how is the situation in Lithuania well in Lithuania we do not have so many smmes and uh startups but nevertheless we have a lot of professionals and who are members of the AI Association of Lithuania and we trying to do everything in our power to empower our echosystem our startups and companies to be also visible in the Europe even we are small country yeah speaking um that that you’re a smaller country in in this sense but how is um how can I say the implementation of AI in the federal government or inmes or in the industry in Lithuania well one of the benefits in Lithuania and small smaller countries that we are flexible and we could communicate with all of the stakeholders quite quickly we could have various initiatives we currently do have the artificial intelligence Forum which is moderating the government uh implementation strategy and it will be provide some guidelines for the economy Ministry of the economy and Innovation and we will see how it will work but we do hope so that it will go smoothly and we will be prepare in really good manner um are there also how can I say universities uh in the uh AI Association of Lithuania is it more industry driven how would you um explain that someone who hasn’t got contact with you yet so since we are host we are really excited that we bring this European air for here in Lithuania for first time and we as ecosystem combine all of the stakeholders the Academia we have the all of the major faculties we covering the all of the major startups and the companies who’s doing Ai and they are member of the AI Association and we have the majority of the professionals who are also joined the the association so in this manner we really cover a lot of and we are umbrella of the AI in Lithuania how is it in in h FR is IT industry Academia and startups at or how is the organization have F structured well we are quite lucky I would say because in the past two years we have uh grow um gerly like it’s uh exponential so we now have startups Academia we have very big companies the three uh biggest banks for example uh so it’s really mixed and we manag to make everyone work together toward um achieving goals so for example uh explaining the impact of generative AI we have been working on sustainable development and I will tell you more about this later today so yes it’s very um mixed and we have this is representative of how it’s going in front right now um just speaking of highlights because um afterwards chlo will talk about sustainability uh Lenas what are your highlights from this session today well we are excited because in our tracks we’ll cover what a act implementation the human rights context will be happening we also have a panel discussion about so how the AI could be used in the Dual applications which is not discussed very often and it’s really challenging since it’s not covered and it is under exception in the AI act and we finally would like to touch the green Ai and the all of the aspects that hey we expanding the computations but it’s have its price so it will be exciting program okay so we are going getting further to this uh Lenas would you be so kind uh to introduce us to our first Speaker yeah so it’s my pleasure to invite for a welcoming talk our vice vice Minister the Erica KK from the economy and Innovation Ministry and she will introduce so what the ministry is doing what initiatives in the all of the efforts for the Lithuanian to be really strong player thank you welcome okay so uh good morning everyone uh it’s really a pleasure today uh to participate and to share some uh fors and directions uh which are working on at the moment Ministry of economy and Innovation uh you know we already before uh the event started was talking about that nowadays it’s natural that everything is about AI I remember at the beginning of this year MIT released its uh uh one of the issues about the uh major Tendencies this year technological Tendencies this year and once again like last year uh one of the major directions of course was artificial intelligence and they sended a very clear message that AI is everywhere even in Sugar what does it mean it means that artificial intelligence definitely is changing not only business processes uh priorities in R&D it also changes our daily routine in general it changes uh the manner how public sector could work and might work so what we are doing at the moment in in in the government in our ministry it already was mentioned uh that we organized uh an AI Forum uh what is it it is a platform a discussion platform which is dedicated to the main stakeholders working in artificial intelligence area it is governmental Representatives it is Academia representatives and of course it is business Community because uh one of the priorities in our ministry is to ensure sure that the business environment for the promotion of innovation culture for the creation of the new businesses uh is is the most important uh for the um modern economy and for the competitive economy so I hope that at the end of uh this forum uh in September we will have a guid lines which we will be able to present uh to the uh major political stakeholders to the high level politicians and we will consider about the renewing of our artificial intelligence strategy I’m really proud that Lithuania was one of the uh actually was the second country in European Union which uh had uh the uh draft and actually uh recognized artificial intelligence strategy among the EU countries it was in uh 2008 19 19 later on in 2020 uh2 we prepared another document like the uh scenarios uh in which directions Lithuania as a country has to go uh in the developing H and in the developing of artificial intelligence so that’s that a strategical dimension next thing which we are also implementing at the moment we are working on the implementation of the artificial intelligence act requirements uh basically it is about the regulation uh it is about the major institutions responsibilities uh and of course artificial intelligence handbooks my priority our government priority our ministry priority is to ensure friendly regulations it is important to regulate because we are working in the legal environment it’s natural but it is not but it is important not to over regulate because we have to ensure that Innovations could flourish in in in the modern economy and modern environment also our Innovation agency is responsible for the establish establishment of uh artificial intelligence sandbox uh as far as I know at the moment on the European level among all EU members we are uh actually the first um country which started to do it last year already we already have the concept of AI sand books we are working um on various Financial incentives we are identifying the necessary infrastructure uh for our AI um startup Community for business Community uh and I hope that next year it will start uh operating already uh and another important thing we’re are talking a lot about regulation we are talking a lot about that institutional organizational structure but it uh but also we have to talk about the skills about the change of the mindset about the change of the mindset among the general Society among the uh public sector employees what I feel as a politician from the general society that they are afraid of a icial intelligence they are still afraid of artificial intelligence because of they do not understand that what about is this technology and we have to work and I’m really happy that the intensity of various events um dedicated not only for the expert Community but also for the general Society is rising so it is important for you as a community to send that message to the people why it is important to understand the value and the benefits which artificial intelligence can bring so my third message is that we are working a lot uh in terms of the change of the mindset and of course the skills so uh once again I’m really really happy that uh we have this format here in Lithuania I’m really happy that our Lithuanian artificial intelligence Association is really active and we are working tightly together uh it is important that the political Direction uh could adapt and could help uh business community and of course academian Community to implement all ideas and all Solutions in as much business friendly environment as it is possible so once again thank you so so much and I wish you an insightful and of course productive day so we are really excited that we have multiple really nice keynote speakers and I would like to welcome our next and the first keynote for the opening speech and opening keynote uh our Thomas lowas who uh Deputy secret Secretary General on the international telecommunications unit thank you very much thank you very much and Deputy Minister ER K dear friends their colleagues indeed it’s a pleasure to welcome you today at this ninth European AI Forum here in villus and is my honor of course to address you on the challenges and opportunities we face today created by the world of AI and a little bit about the governance to ensure that thei is work for our sustainable future so indeed a year and a half since chbd captured our attention and headlines worldwide a has become really the very familiar part of the lives of millions of people yeah driven misinformation and potential job losses being weighted against unprecedented process efficiencies analytical cloud and simple convenience of everyday tasks however crucial AI governance discussions are moving from high level principles now to the Practical realities of deciding how we should or shouldn’t be able to use AI so we just three weeks ago we had our AI for good Summit in Geneva which featured actually AI driven breakthroughs for people on planet and a few of them I would like to mention for starters in healthcare for example groundbreaking brain machine interface Technologies which are poised to re revolutionized neurology and clinical treatments AI driven surface electrography which is a technology that leverages electrical activity generated by muscles using service electrodes and gives patients better than ever motor control Innovations of Prosthetics and Rehabilitation engineering which unlock bionic dxterity and indeed was very a little bit of emotional actually moment on the stage where an LS patient Louise from Portugal who lost his ability to speak was able to communicate with a with the audience using the uh the company startups enables Hall device and AI Technologies as well indeed existing AI Technologies address all the 17 SG goals they enhance Earth observation satellite data climate monitoring early warning systems and growing need in the harsh phase of climate change our AI for good Innovation Factory which we had launched or actually awarded winess at cop 28 last year highlighted a number of startups working on groundbreaking um Technologies to leverage AI all over the world for example company like padon reducing fish feed costs tlby from sagol utilizing satellite data to increase cashew fields in West Africa and others AI is also helping nuclear energy it use an international atomic energy agency’s challenge Innovation Challenge in have of Fusion Energy is simulating Fusion to avoid radioactive waste cut engage AI Technologies also support peace building and humanitarian operations worldwide and then AI also AIDS in LA enforcement analysis for example helping save or protect children from harmful situations and all these different applications indeed have been explored in AI for good platform so which is powered by and supported by 40 un Partners this platform features our annual Summits as well as extending yearo program of online events as well as continuous Community engagement through our new roal network platform which now has around 26,000 participants this year is the a good Summit which will took place in Geneva next next to Wi plus 20 Forum high level event also so it attracted 8,000 participants 6,500 of them in in person and scoring how much AI has captured public imagination and indeed the crowds that were conven you know they were queuing Gap in front of the conference center there with a very visual Testament to that so it’s the for good Summit started 7 years ago as a solution Summit but of course now has worked into GL Global showcase for responsible AI development and indeed important word these days is responsible when we talk about AI development because the new reality of a powered misinformation disinformation and Def fakes is rather unsettling especially in the year when around half of the world’s population are going to the polls gender racial and other biases are also deeply perpetuated by insensitivity trained AI models and a recent study Global study showed that around 45% of AI models demonstrate gender B which are direct impacts on different aspects from Health Care to employment opportunities we also cannot disregard AI use in the Warfare harmful aggregating uh harmful aggravating social oppressive or lethal users only stroke growing fears of AI as people imagine a future dominated by autonomous machines so real risks from AI need to be recognized and they need to be addressed however just focusing on risks also undermines huge potential of a for example the studies show that the could add around $4.4 trillion US dollars to the global economy at the same time digital Solutions including AI can help meet around 70% of sustainable development goals as well and clearly therefore we need power fi harnessed but harnessed responsibly safely and sustainably so appropriate governance is key and already some of the instruments were mentioned Europe just a months ago actually featured a few achievements of course one of them being you final adoption European Union AI act but another one which I think we’ll hear a little bit more next is Council of Europe’s Freeman convention on AI which is the first ever International treaty in that area other jurisdictions such us China as well as others also developing their own AI regulations and we have a number of international processes such as G7 Hiroshima process as well as AI safety Summit process with the last Summit in So So Korea just a month ago leading policy makers and coers of these initiatives also came to Geneva on the 29th of May in the first ever hour AI governance day which was part of AI Summit it features around 200 participants out of them around 70 ministers you know 25 un Partners around 100 Academia and Industry people and more than half of them came from developing countries and Minister for Bangladesh then spoke for I think for many of them when he noted the the absence of many countries developing countries from the global AI governance processes and he really noted importance and emphasized that the countries everyone needs to be included in those governance discussions policy leaders there emphasize a few points that they felt are important when as we design the future Fair governance first is responsible Frameworks so tying AI closely to ethics and human rights second is interoperability inability of both platforms technical interoperability but also interoperability in terms of policy framework so ensure that our policy and regula environments are compatible and this interoperability should be underpin by third international technical standards leveraging they also noted the importance of leveraging AI to bridge divides and not to create new ones as well as emphasize Global need of global solidarity and resource sharing to make sure that the AI doesn’t leave anyone behind and indeed one that that is an important exp of the divides and the risk of having a new dig type of digital divide is AI divide F indeed just a reminder still 2.6 billion people worldwide are not connected to the global internet so they are left out of the global digital Revolution let no loan of a revolution and of course developing developed countries host around 80% of collocation data centers with one of the country actually having around 40% of them which leaves the rest of the world to share the rather small number of them the Divide is also evident in a number of other areas such as Innovations and patents concentration some studies show that L 50% of patents are concentrated in three countries countries different place in AI value chains some of them producing key components such as microchips or employee Engineers to develop foundational models where this is as a source of raw materials or data uh data labeling and data sorting as well as the policy divide so our survey of our memb state shows that around 85% of the countries still lack AI regulations and more than half of them doesn’t even have ai policy so we and when I talk we as un are responding and we are responding to make sure that as in the words of Secretary General Antonio gues just said just a week ago in the at Council that AI doesn’t stand for advancing inequality indeed for several years now we’ve been un has been having a set of activities in this regard our recently launched un system white paper and our governance features around 50 normative acts both directly applicable to AI around half of them as well as closely related areas that are applicable to AI as well by the UN system they include overarching broad guidelines such as un ethics framework AI ethics framework as well as sector specific ones such as guidelines from W or guidelines from UN agency called unicri on use of AI in law enforcement this inter agency working group started in 2020 uh it brings un family together and I have a uh have a pleasure to co-share that with the UNESCO assistant director General Gabriel Ramos and this of course this work through this through this work it contributes and complement high level advisory body on AI set up by secret General that brings external experts to help un figure out how to govern artificial intelligence un agencies also Implement AI projects actually we launch now annual reports on una activities our recent one demonstrates around just over 400 projects by 47 agencies that leverage AI we also have different uh different collaborations on different cation areas such as our collaboration on AI and health with who and wipo our collaboration uh with wmo World Meteorological organization United Nations environmental program on AI and disaster management and so on we also as at have already have developed or in the process of developing around 200 AI related standards overall so all in all our work we also make sure that b Place strong emphasis of including all countries both developed and developing alike so for example I our challenges in machine learning and we had them around 80 so far involved around 8,000 participants on around the world and those who doesn’t who don’t have access to compute resources will actually provide them for free at the recent AI for good Summit launched a few other initiatives to support AI governance worldwide first it’s launched the unified framework for AI standards development Under the Umbrella of world standards corporation that brings together itu together with International transation Organization ISO and international electri technical commission I second we launch a collaboration for multimedia authenticity AO water Ai watermarking and de fake detection with a number of private and public sector Partners to tackle that very specific challenge in in the world today the third we launched AI for good impact initiative to mobilize how diverse active stolder multistakeholder Community to to share knowledge and assist developing countries undp and UNESCO agreed to collaborate on Country AI Readiness assessments to help countries to be better prepared for that and finally we together United with United Nations University unu started the process for the flagship AI for good report that will transform the AI for good platform knowledge in the uh valuable resources stakeholders one aspect I also would like to mention mentioned I I’m very glad it was mentioned opening is sustainability and AI environmental impact indeed just a reminder May this year may become became 12 months in a row hitting the record high temperatures worldwide and as AI climate crisis already steers us into face we must decide whether AI is part of the problem or part of the solution and digital Technologies of course including AI can Inc improve Energy Efficiency optimize inventory management enhance business operations reduce emissions and new ways across different sectors and actually it’s by some assessments and can mitigate worldwide around 5 to 10% greenhouse gas emissions by 2030 which is equivalent to Total emissions of the European Union so and of course AI can also help climate predict climate and weather patterns guiding and helping early warning uh Warning Systems preventing us from disasters or helping us to mitigate disasters but the tech sector itself also produced 1.7 at least 1.7% on greenhouse gas emissions and even though overall emissions for the tech and Technology sector are rather stable over the last years AI represents growing very fast growing emissions proportion to that actually emissions from data centers cloud data centers between 2020 22 increased by 45% and energy consumption 57% AI is also very thirsty you know actually 10 5 to 10 promps consumes around half a liter of water so every time you send 10 promps on sht think of how you just drank the bottle small bottle of water so indeed starting last year with the last year’s climate change conference C 28 it has brought together Partners around the world Under the Umbrella of green digital action to Galvanize digital Industries action to take responsibility for our actions in in the in this face of the climate crisis and indeed these forward looking companies have embrace the target of 1.5% for sorry 1.5 degree limit for global warming and they all together asked to share their own but also other invited other companies greenhouse gas emissions data at the public database we also together aim to ensure the sustainability standards are implemented not just adopted and we really invite everyone to join us on board in that in that journey and of course we’re looking forward to cop 29 in Baku this year as the next milestone in that regard so ladies and gentlemen we all have responsibility to make sure that the AG of AI become the age of prosperity sustainability and inclusion not the age of the fear anxiety or Division and to achieve that we need to manage the risks of eii ensure that these benefits available to all and also include the whole world in shaping the ey future thank you much and have a great event thank you thank you so much thank you too much we really appreciate your input I learned that AI is thirsty uh this is also something to remember by and um very um very grateful for your for your insights on this part so ladies and gentlemen we are now coming to the Netflix part of this event uh we have a video message um um from and we’re very happy and proud that the German Federal Minister for transportation and digital Affairs Dr fulker wizzing is um having a short speech to us via video and now we press play do we yeah esteemed European AI Forum participants distinguished guests from the business and scientific communities from politics and Society ladies and gentlemen our aim is artificial intelligence made in Europe the best way to reach it is for many stakeholders to get involved contribute communicate and connect like you had doing at this conference and Beyond you are driving Ai and AI is driving technological progress we want to tap the tremendous opportunities AI offers this means that AI developers have to be given the freedom and flexibility they need to try things out and experiment so much is happening in Ai and at such a high speed no one here knows the limits of what AI can do today today or will be able to do in the future that is why it is imperative that AI regulations are adaptable and can be updated as required for example at G7 level we successfully championed common guard rails for the development of advanced AI systems this code of conduct is a true milestone for safe use of AI worldwide it helps to ensure that AI is developed in line with our value this strengthens people’s trust in AI applications this is also what we want the European AI act to do we worked to find Innovation friendly rules during the consultations the compromise we reached provides a reliable framework and guidance now it is up to us to implement the AI act our implementation should build on existem structures without losing sight for our needs data is the foundation of AI Innovation it must be widely available and easy to find access and use we can and must improve this in Europe much of the data that has already been collected remains untapped in the future the European data act will help change this and we are already looking into what else is needed for example through our Innovation Club Estonia Latvia Lithuania and Germany have joined forces in this club to boost Europe’s standing in the digital sector and Champion Innovation friendly framework conditions to this end we have drawn up a list of Demands for an opportunity driven digital agenda for the next European commission our aims include cutting red tape removing unnecessary reporting duties and breaking down Market barriers we also want to grow the AI and data economy the the end goal is a truly complete digital single Market in Europe This is a challenging task that policy makers and the administration cannot shoulder on their own that is why we need you the stakeholders from the business and scientific communities and Civil Society we are counting on your impetus and vision as well as your optimism and passion for digitalization we need more of this very mindset across the board to drive digital change now I would like to wish you every success for your event interesting conversations and lots of new insights and inspiration I thank you very much and wish you all the best so it was really nice to have a welcoming message from the minister and now since we having the presidency of Lithuania in the Council of Europe in this half a year we are really glad to invite the chair of the AI Committee in the Council of Europe the Thomas Schneider if I just say next slide can you then move it or do I should I do it myself okay so I guess it’s this one one okay so um I will say a few words about also how the Council of Europe is trying to contribute to uh safe Innovation as we call it so how to regulate AI uh while fostering Innovation and um so now I have two hands and three things let’s see how that goes so first of all I think we should not forget that AI is not the first disruptive technology that we are facing and it can be the way to deal with AI or the effects of AI can be many ways be compared with the invention of engines in the 19th century because engines were either put in vehicles of all kinds to move people or Goods around or were putting machines to produce food or other things and they were replacing human labor or animal labor like horses and so on and something similar is happening with ai ai is used to uh create content or process content or process data uh which is something that used to be done basically by human beings so it does not replace uh manual labor or or physical labor but actually cognitive labor and if we talk about AI I think we should also talk about platforms because this is one of the key applications not the only one uh of AI and platforms are an extremely powerful tool to bring together um supply and demand you can have a personalized demand uh and get a a personalized offer and this is something that is uh extremely more efficient that all the traditional ways to bring together uh offer and demand so far so platforms will make their way we we already know it they started with booking.com and other platforms now we have them all over the place so they will replace lots of other uh uh ways of uh bringing back uh bring together offer and demand and uh I actually can see it here so that helps and of course we also know about the many uh risks and challenges uh that AI bring uh with it of course people are afraid to lose their job that is one element that we’ve heard we are afraid to uh uh about the functioning of our democracies manipulation we are afraid about privacy human rights we are afraid about being dependent of big Tech Gatekeepers and so on and so forth the core fear is basically for people individuals for whole societies for for industries that we lose control over our lives to machines or probably more probable to people that run the machines or have control over the machines if it’s not us and U so there is a need everybody says yes we need to regulate then the question of course is how and one challenge uh in the be that we had in the beginning was like what are we talking about what do we mean when you talk about AI uh fortunately uh the oecd has helped with this and they have agreed on a definition that also the Council of Europe has started to use and that has made its way so for the time being at least for legal purposes there is something like a accepted definition of AI then of course what makes it difficult but it was the same with the engines is the risks and the opportunity you use it depends on what you use it for with what data and so on so there’s no one size fits all solution and the other thing we’ve heard it technology is developing really fast how do we manage to govern a technology uh by written laws which is changing so fast which even the ones that program don’t really know exactly what it’s doing how do we live with the fact that we are National jurisdictions or in the case of the EU a supranational jurisdiction but it’s a global phenomenon we buy uh services from anywhere they are developed somewhere and then used elsewhere one of the big questions um that we have in my country for instance is should we go for horizontal legislation like the EU does with the AI act which is of course also accompanied by other by other acts and other tools or should we go for a sector specific regulation there are many ways to roam also the question of what should we do in terms of binding laws how much can we do with Co and self- Regulatory regimes lots of questions but then again let’s let’s look at what is actually really new what is really different where theyi compared to other Technologies other ways and if we actually look at our existing regulation our existing regimes from human rights to to to standards the Gap is actually not that big as as some people think so most of the situations are basically also covered when it comes to AI but there is a gap there is a need to clarify a few aspects and also there we can actually learn from how we dealt with engines in the last 200 years we don’t have one engine law one Council of Europe convention on engines that solves everything we don’t have a Ministry for engines we have thousands of legal Norms of technical standards and of social cultural norms and institutional settings that together somehow help us to deal with engines and there is a different level of harmonization depending on the context so if you take the airline business um it’s more or less globally standardized how to land the plane on an airport what the words are used and so on if you look at the at the car system uh the Germans still uh have some motorways where they have no speed limits in other countries it is different the people in the UK drive on the other the side of the road but they can’t take their cars to Europe they just need to pay attention and in particular the other ones need to pay attention when the brds are on our road so there’s different levels of harmonization again based on on the context and we’ll probably will need to do the same for AI because social cultural differences in particular on how to deal with risks what you defer to the state to protect you whereas in other cultures you rather build on the self uh uh on the awareness of the people to to take care of themselves so that will not disappear at least not quickly so we will have uh some harmonization in some areas but less in others and of course uh the comparison with the engine is something I really like but there are differences it’s not so easy to copy uh an engine or to copy oil or move oil from one place to the next whereas of course with digital with digital uh bits you can move them around and copy them uh much faster so we’ll probably have more globalization we probably need more International harmonization than we had with with engines so of course we are actually for quite some time uh if you look at the international institutions and we had Thomas lowas from the itu they started with the AF a good Summit in 2017 that was about the time where actually all of the international institutions started to work on AI we do have some soft law standards like the oecd that was one of the first with a recommendation that was just recently updated you have the Council of Europe with sectorial uh guidance on on on the media system on human rights on on in the judicial system then of course the first Global instrument was the UNESCO recommendation on ethics of AI and then you have tons of guidelines principles already before by industry players by companies by company industry associations by think tanks and so on and what is missing here is we have of course technical standards that are being developed at the itu at the iso at the i i e and so on and so forth and there is more and more cooperation between these and now this is like what is happening on National levels and um you probably know this famous pyramid over there uh since uh this is a country of the EU so we have the EU AI act uh which is one way to try and regulate AI in a national or supranational jurisdiction based on this famous risk-based approach um it it is interesting to hear to hear uh the German Minister wiing because uh everybody says that they all try to do the same thing we want to innovate we want to Foster Innovation but we want to protect human rights and societies and and and the functioning of our democracies but there are quite different ways to do it and of course everybody thinks that they have the best and most Innovative and most agile solution time will tell um so we have the the EU approach then the US has taken a different path based on voluntary commitments on self self regulation uh complemented with an executive order of a few months ago that outlines a number of uh policy goals in particular targeting the the administration itself less the private sector with a number of measures the UK is choosing a similar path they are not going for a horizontal legislation they try to adapt the sectors boost all their Regulators connect them Empower them um and and have like a central Institute that is supposed to help with the empowerment Japan is going in a similar Direction my country is also thinking about how let’s find something in between the AI act and and what the UK and others are doing because of course for us it’s clear that we need to be interoperable with the EU because we’re in the middle of Europe and our economies is so connected but we may be trying to go for something a little less complicated uh from a Swiss simple perspective but we’ll see uh where where where we end up but again the goals are normally the same but the ways are different and will not have these things fully harmonized because these things also depend on the history of the institutions there’s reasons why the EU did it the way they did because they have 27 member states they want to have a harmonized single market so there are reasons why the US chose their way because they knew they would never get something like the AI act through their own Parliament and so on and so forth so the question is how do we achieve the operability interoperability that we need in order to actually be able to function across borders and exchange services and this is where the Council of Europe my view can contribute also Council of Europe has started to work with a on AI as I said before 201617 and then spend three years studying what are the gaps what is actually needed in terms of legal instrument on human rights democracy and rule of law for three years and based on that then the committee on AI that I’m I’m chairing uh since since April 22 got the mandate to to develop a legal framework that would be based on the Council of Europe standards on human rights democracy and rule of law so it’s not a market regulatory instrument it’s a human rights democracy and rule of law instrument but it should be conducive to Innovation so again we see the same goals all over the place plus and this is important because these are all papers uh we need something to actually operationalize these Concepts so uh what we uh will develop until the end of the year is a so-called HUD area a human rights democracy and rule of law impact assessment which is something similar that what you’ve seen with the risk pyramid of the EU but it’s focusing purely on on human rights democracy and rule of law and not on things like product safety and others because and so it should be complimentary to what the EU and other standardization uh bodies are developing and uh it was a very intense process uh that led us to the end of the negotiations this March um again the purpose was if there was no consensus to create new rights or strengthen rights but at least we wanted to keep the existing protection levels for democracy human rights and rule of law and it was clear to many that it would maybe not be enough to just have a European instrument because we have the court of Human Rights we have the EU AI act that will have an effect also on countries like mine and others but we felt it’s important it’s the Council of Europe is the only institution that may bring um uh countries like-minded countries that share the same values together together in a more or less Global instrument and we already had 11 countries we have all the G7 here if if you look at it that were negotiating and they all signed up to this and it was an inclusive process with civil society and and others on board and so the general logic of the of the Council of Europe um convention is it should limit itself to fundamental principles that are complementary to existing uh human rights instruments but also to existing other instrument and should just fill the legal gaps that the new technology like AI poses it should be future proof so it should not go too much into detail because the the the convention on data protection lasted for 40 years until it was revised so that is the the challenge so it should actually at least last for like 20 years and the biggest fight was of course the scope of application because countries like the us would have preferred that this is not applicable to private sector just the public sector but the expectation in Europe was uh quite clear that it should also cover in the end the compromise that was found was it covers all so states are obliged to address risks by All actors but it gives them a great leeway a great choice of what measures so it can be uh legal measures it can be self-regulation and so on and this is basically the the shared consensus that also should help us becoming interoperable between these different uh legal systems these are the principles I will not read them out loud uh the most important in this area is the others are human rights principle the last one is called safe Innovation so it should also Foster Innovation uh Thomas said responsible Innovation it is here called safe Innovation but it’s basically the same idea and the AI act also has a Provisions about how to create sandboxes and and other tools that should help uh allowed to innovate and be safe finally a few remarks looking ahead we really think that the Council of Europe framber convention is an important tool to safeguard human rights democracy and rule of law on the one hand but also to to help like-minded countries come together make sure that they cooperate with each other that they try and keep their systems that their logic’s interoperable um and of course it is clear that this is not again not the solution to everything it’s one Milestone but there will be uh will need to develop again hundreds or thousands of technical legal and social cultural norms in order to deal with AI in the different contexts and it also offers a path for cooperation between governments also with ones that may not yet have signed or ratified the convention so it provides like the cyber crime convention did successfully too it provides for a framework for cooperation also with countries that may not be close enough to to sign up to it but that may be willing to cooperate W with us and the final the final point is this and this is a general remark if you take my country we normally take like five years to develop a law and then we have a direct democracy so if enough people don’t like it they go for a referendum so it takes a year that think people will vote on it and if the majority of the people don’t like it then there’s a no that means back to square one it takes another five years to develop the next the next version so we can take from five to 10 years to develop a law uh the technology will probably slightly evolve and the challenges will slightly evolve so I think that we really need to also adapt our governance models and mechanisms to new technologies we cannot work with the structures that were actually created or influenced by the first Industrial Revolution our uh in the mid in the middle of the 19th century where trains were uh starting to run across the continent there was no France there was no Germany there was no Italy these National States the parliament didn’t exist but also the Industrial Revolution led to the emergence of M you had the entrepreneur party you had the workingclass party and then you had the farmers and we are still basically uh using the 19th century governance tools to try and deal with the challenges and the technology of the 21st century and we are realizing that this works less and less well that we are just not able to represent the people with these Mur that the parties are breaking apart but also not to find a solution as quick enough so we may have to rethink we may have to use AI also to develop legislation or to develop standards it’s already actually it’s already being used and we also may have to think how to re think our parliaments are ways to make law and and maybe also that they change dynamically not just every five years but they they change almost daily like the technology develops thank you very much for your attention question so uh no please take the microphone thank you so much for for your very engaging um keynote here so let’s start at the end of your uh of your keynote that Parliament and governance has to change I have how can I say a life motto which is called Happiness is the right expectations right expectation management is is do you see tendencies that government governance changes because I think with the UA act we saw that it is still in the 19th or in the beginning of the 20th century well there’s one thing and Thomas lanowa knows it very well on global level you have a big discussion about multilateral processes where so-called multi-stakeholder processes where you involve not just representatives from governments or parliaments but actually the industry the Academia the technical community civil society and I think this is a way to try and have a more representative uh Gathering of people also more expertise because if you look at our parliamentarians or our politicians how much do they know about AI so there are people that decide about technologies that basically they yeah yeah are not really experts let’s put it diplomatically and this makes only limited sense of course they are uh elected by the people so they have a legitimacy but they have no knowledge and then the question is how do you Empower them or would it make sense to to also there be more agile and get the people at the table that actually know about the issues also more dynamically dependent on what the challenge is dependent on what the issues are not having to go through the Gaye keepers of parliamentarians but then you need to find the regime to to have a legitimacy for a new model and this is what people that don’t like the multistakeholder approach say well governments are elected so I represent my country my people it is more or less true depending on your democratic system and the stakeholders are not elected so who are you to speak and we are trying to to to find ways to actually provide for more inclusive more agile more accountable and transparent policymaking structures on International level also on National level we are a bottomup democracy so we have to be inclusive otherwise if you ignore people they just say no and then again you go back back to square one I think we it starts by being more inclusive more transparent more accountable and then we need to find ways to uh redefine representativity maybe less through geographically elected party Representatives but through other skills or representativity of a particular industry of a particular knowledge of a particular interest but not necessarily through the the party logic from the 19th century when you said um when you showed all the guidelines all the International papers uh which are surrounding AI at the moment I personally believe you can train a large language model on all the regulations and and and and guidelines on the national level um on this how can you harmonize this because I think sometimes every country has their interest of doing their own strategy of doing their own recommendations and even in the adaptation of the AI act I have the fear that we have one AI act but 27 different views on it how would you harmonize that again I think it never will never be possible to harmonize everything to 100% it is probably also not really needed and does not make sense because people are different people’s life are to some extent different we are assimilating ital is not that much different anymore from Germany like it used to be 30 years ago when it came to to rules but um there there are technical standards that will some of them will make it others won’t and the ones that make it they will harmonize from a technical point of view maybe product safety what else you whatever you can solve a technical standards there may be two competing standards we used to have if you the older ones like uh the video uh standard in the US was different so if you wear in us and bought a VHS cassette or like great and then you put it in your thing at home it’s like didn’t really work so we’ll probably want to avoid this but also on the legal basis like this is the EU will have a an effect on other countri like mine the UK uh they say well we are not really looking at what the EU does but of course they are which is the narrative um and the Council of Europe convention will harmonize it from from a human rights perspective uh and the good thing is like the ones that are members of the Council of Europe they Bound by the human rights treaties anyway so they they basically have to follow it anyways but we have all the G7 we have a number of we have Canada we have a number of Latin American countries so there will be some harmonizing effect through this uh on the legal basis but then the social cultural differences will probably remain so yeah and and this is a trial and error again we we are after 200 years of engines we managed to bring down the that people in in factories de people in car traffic that people airplanes are not flying falling from the sky daily but we didn’t manage to do something against the biggest challenge which is the CO2 problem the the climate change problem after 200 years well we didn’t really know about it for 200 years but we know about it for I don’t know 50 years and we haven’t managed to solve that problem and it will be probably the same wherever there’s an interest and the lwh hanging fruits will be caught very easily but the bigger problems it depends on whether there’s a a common interest to actually work together and solve the problems so let’s hope that this will be the case to much everyone thank you thank you so ladies and gentlemen up to the next point is of course at the moment we are discussing the the implications of AI and and the problematics um that we’re facing and certainly the point of disinformation and automatic influence is a challenge for society economy and democracy and we are very very happy from the uh Professor vitval um to have you here for your keynote we really appreciate please a little Applause thank you very much well uh what I will tell you is um showing some developments caused by AI also now speeding up like the technology developments and for this I will give you other view angle on things you already know so because I’m a professor I will tell you a little bit about history uh so about history so uh how did AI begin so it began with with pattern recognition in the mid of the last century so I have had an device that I would like to have for pattern recognition and I needed a trainer who trained the system so by this um when there was an input the trainer said this is a right pattern this is a wrong pattern and after a long training he has implement this in a system that was able to uh recognize a dedicated pattern we have now three technology trends that lead into uh let’s say AI as you know it nowadays so one that started a few decades ago was the improved Hardware so in the beginning it means I was able not to do one pattern recognition but many pattern recognition in the same time so in this example if I have the input of square I um I couldn’t only um dedicate detect the shape but also the color so I was able to create systems that have been able to uh to recognize both patterns in one the next development that leads into AI has been the so-called self-training algorithms you know it as machine learning deep learning however you name it what happened there was that we have added a kind of feedback loop within the uh pattern recognition and by this we have optimized the uh probability of the recognized pattern and as a result the system that was trained or is still training itself is able also to recognize patterns that it was not trained on the third development that has made an impact is this huge amount of training data that is available nowadays and the impact of this is that we are able to uh make the probabilities of the pattern recognition more precise and by this we have now done by the global tech training of models by all available texts by all available pictures by all available audios by all available videos and by all available codes so this is roughly the development into uh Ai and the question is what kind of principle um applications are created with this so coming from a technical point of view um and then I application is a kind of response creation system so I have an input I have this pattern recognition and then I create an output and from the beginning it was for identification so via an i AI system I’m able to uh ident uh identify for instance the meaning of a text or um I’m able to identify by Audio a music the song or the style of the music maybe also the singer of the music I can also um identify voices or the phone of the voice so all of this is feasible by the pattern recognition or by AI but also the uh identification of sceneries on a picture or the persons on the picture and this of course you already know um if I use the response also for creation that I just try to check which kind of pattern fit fits B you you are already aware of the prompting so I prompt something as an input and then the system creates something as an output so like you know from chat GPT text can be you text prom can be used to create text for instance for chats or for translations or summaries and all the applications that pops up during the last two years with audio I can do the same I can use an audio uh and prompt to create music so the Music Creation is now as professional as you do it as a human composer you can create Now voice clones where you can give your voice to any language uh to any kind of uh singing so this is feasible nowadays with prompting and also the prompting of images of videos is nowadays feasible I saw the first trailers of movies that have been completely prompted by by AI this week so it’s it’s really incredible if you now just switch the words input and output to action and reaction you come to the next application of AI systems and uh this is a control Loop so a control Loop technically is something you want you have a monitored system and you want to to keep it on a dedicated measure and for this you have an imput for instance a square and our control measure is a triangle so the output should be a triangle and um the question is now what is the most interesting system to monitor well you are the most Wanted training data set in the world so you are the monitored system and if we put uh let’s say controller below with the measure to keep you happy if there’s an input for instance uh yellow Square then you become happy so far the theory to make this happen of course you need a very complicated uh pattern recognition system with very sharp uh propagations and propability uh sets that is self-learning and basically the foundation for this is already given uh you can see this in the slide from uh Christy wer from crack labs this is already 7 years old and in this um uh overview you see what are the data inputs that you give give to the system for free so every interaction you do on the Internet or on digital systems like your clickings like your swiping like your video views your registration on Services your app installations your likes your moves that you do when you have your mobile phone with you the content you share the posts you take the purchase you do and even the call center calls they are all going into a kind of system where different companies process all your datas and this affect what you get so what you see here is mainly the feedback loop that I have shown you before and what you see is and you already already know from your experience that this will lead into the content that is shown to you that leads into the products that are recommended to you and it’s also the way how companies try to uh shape your behavior and we know also that’s not only for marketing purpose but also to influence elections like the Cambridge analytics Scandal has shown a few days years ago so well we have now correct now the feedback loop a bit because the most important control measure for economy is not your happiness but money revenue and profit so understanding that Ai and petang recognition is used as a kind of control Loop AI becomes now automated influence so this pattern recognition from the last century with a huge am out of data in the hardware enables Us Now to create this automated influence and the monitored system of this automated influence is now a dedicated group of persons and the control Loop may try to get them all happy or may them eye to buy some nice products or to elect a dedicated candidate on the next election so by this automated influence I’m able to NCH complete Target groups what I would like to give you as a discussion starting point is a kind of impact map very rough of this automated influence and I have tried to give it a bit Direction into more bright application and more dark application and the impact on economy society and democracy it’s not complete it’s just a starting point for discussion together so let’s start with marketing marketing is not that bad but also not that bright and at least is the at the moment legal automated influence we have what we see now coming is an automated content generation and this has uh free control measures or measures why we are doing that one is that we want to uh control this decisions for instance for marketing or for uh for elections the other is Attention our attention for instance to stay on social media platforms or to stay on steaming uh streaming um Service uh platforms or our excitement well in example if we are online Gamers that we still continue gaming or against steaming uh Services where we still uh watch the movies so so what I mentioned already is the entertainment part that has of course an impact to society and if we think about this automated uh content generation then of course the brightest impact would be education if we use the measure to make young and old people Consciousness more self-aware and so on what we also see is that by uh automated content creation we get disinformation and thisinformation is if you like The Unwanted automated influence and The Unwanted automated influence has the aim to descise you or to harm you and there we have seen different examples um so for instance the the number of shock calls by voice clones where somebody is calling you as your daughter or your son with the voice of them to send money somewhere because there was an accident or something like this is increasing I personally heard it now a few times during the last months from my personal um environment from people who have had this kind of calls then you have also the manipulation by realtime deep fakes in 2022 the mayor of Berlin um Miss gifi was uh was talking to a fake mayor of Kiev Klitschko luckily it was just a comedian but um that was the first public event where realtime deep fake was used on a politician and what happened this year in February was that there was um company in Hong Kong an international company where happened a video call and within this video call there was an employee of the finance department some middle managers and free members of the board of management who have been Complete deep fake realtime avatars and these three guys told the uh Financial uh Department that they should transfer 24 million US dollar to some um bank accounts this happens and this was the first huge bank robbery or uh robbery if you like done by Deep fake uh this year and from this case we can assume that there will happen even more like this in the coming future because the technology is improving and because uh the systems become more user friendly so my big hope is that we can create by the regulation and the things we have heard also uh before on our European commission that we create together something that is doing a bright future for democracy because AI could have their an impact to from my point of view but I have no examples at the moment we have to create them together maybe we start nowadays what I can see is that we get into something that is something like all world’s 1984. what does it mean so we know that we can create deep FES quite easily but we can create any kind of uh deep fake news without any problem anymore we can we can prompt already compa campaigns with the content of the meme like the the photos like the text released to this and and so on and we even can create now fake sources like the pink slime movement in um in us where we make it also impossible for fact Checkers in future to check is this news real or is it a fake because I could create by AI immediately a website as a local newspaper where I just mix real news with fake news and thinking about this that you can do this by Ai and you can do it in real time this is the um extension of the information or of the ministry of information from George aille where you can instantly change the history of the whole world at least in the digital fair so well for the future there are three attractors that I would like to show you well if we look at the next 12 to 40 24 month from now we see already some developments you know them I have explained some of them like um realtime translation the real-time deep fakes The Voice clones but also all this uh act like Bots act like a psychologist act like a lawyer act like and so on and so on what we see now from the last example I described with the fake news that we will get a loss of trust in digital news and digital sources why because most young people people and people inform themselves via social media platforms and fact-based content will become uh available only behind pay walls that means that uh many news of fact-based information are not available for the mes anymore and because we all knew that the AI generated content on the internet will increase um it becomes harder to find really trustful information in future so you see this already slowly coming as an attractor the second attractor is the loss of trust in digital communication I mentioned the example of the um of this voice clone of the shock calls I mentioned the example with the real time deep fakes and now imagine that um that parties use this or proxies uh use this to make speeches from CEOs or from politicians um or that this happen with the communication of yourself so if you are not sure if the phone call you will receive or the video phone you you will have is with your relatives or if is with a AI or is with somebody else so that could make a real huge impact if this is uh Rising faster than I try to Hope or I hope at the moment um the last thing I would like to mention is uh the loss of trust in privacy because if all your data is a training data set for pattern um then you are completely naked to the public or at least to the to the algorithms and the other thing is um if you’re not on the social media platforms somebody could create a profile of you that looks like you talks like you act like you and harm you by this because this profile can make statements and also disturb your whole private uh private life so and if you then also know that we have this emotion recognition um speed him up and you combine this with the other pattern recognition then you have an first idea what the current Technologies made an impact on society as a conclusion I have no conclusions but I have some recommendations from my point of view on a general level ask yourself how much economy-driven AI is healthy for our society and our freedom second start a common development for bright usage of AI to benefit our democracy on a social or on a society level I would say let’s raise data and eii literacy and I mean automated influence not only um artificial intelligence second establish s sustainable trusted news sources and third create secure fallback communication Solutions instead that we lose trust in digital communication on a personal level I would like to recommend you become data economical use free opensource software and have S words in case of digital emergency with all your beloved and related thank you thank you so much um well that was of course perhaps realistic but also perhaps oblique outlook on that just for the sake of discussion if I can challenge you on this um doesn’t every new technology which comes into a society has this light and dark risk so so when you take for example um um the the emergence of movies and then came leny reeny and did propaganda movies uh for uh in in in Germany at at the Nazi time um isn’t that always an inherent point of progress this light and dark thing uh yes and there is no right and wrong um just because you mentioned the movies uh I was uh attending a workshop for the future of the movie production in bburg last year and we thought about the impact of AI and there is a statement of Marvel Studios from last year that within the next three years so next year they have the com First Complete produced AI I produced a super hero movie on the cinemas MH and we have then thought what does it mean for the production and we have tried to just just think ahead and if we think ahead what will come if all the systems come together is that you will get an realtime movie making that is individualized to your personal demands means we both like Star Wars but we do but the point is we see the same movie but in your movie Prince Princess Leia has blonde hairs in my she has dark hairs the fighting scenes for you are very bloody with all all the things in my in my ones you don’t see it you just see the idea of a fight so you see this kind of individualization and customizing of movies will go ahead what means but this is fantastic yeah and we are then yeah and then you are addicted and you will not lose it anymore so still icted but no no but but but I get what you’re transferring here but isn’t the just also for for challenging of the challenging state is there and I had also discussions with the creative industry with the producent alliance and and sometimes creative Industries in in my thinking are quite conservative in changing the business models but isn’t there um a a constant development of um how the work is done who is engaged when you see the the transformation from animated movies to CGI movies and stuff like that well uh you are right but I have spoken with the guys also with the um uh heading technology guys and the point is this the development of the Technologies is forced by AI is so fast that even if you try to to keep it for your daily work you can’t do it so if you are not willing to change it’s harder but even if you are willing to change the speed is so high you can’t change for instance synchron speaker for German or other movies you don’t need it anymore because you have instance translation and you can use any voice clone so why don’t you need this this actors anymore it’s just one example yeah yeah no no most definitely I I agree with that but um however in one of my um favorite bars in Berlin they make silent movie nights and there is a big poster there um which says uh stop the movie with sound and that was the association of Orchestra player in cinemas and I think that this is the same development what you can see here and you can’t stop it also this actor protest from last year in Hollywood yeah they can protest and you can they can protect their rights on their person but because of all this pattern recognition AI can create a synthetic actor that is more fitting to your demand and my demand than the original one so we can come around all this copyright issues by this and this will also create an impact so yeah sorry for that no no problem F VF from the it thank you so much so ladies and gentlemen the next content I think is not AI generated it is a video statement from AI verier the Ambassador for digital affairs from the ministry for Europe and foreign affairs from France ladies and gentlemen I would like to thank the artificial intelligence Association of Lithuania for organizing this AI forum and this round table on the implementation of the AI act with with respect to Human Rights I have not made it to Venus but I am thrilled to contribute to this discussion through this recorded video I hope that underlining the French attachment to the protection of human rights on the digital space through commitment to the European AI regulation will contribute to the discussion positively today AI systems are everywhere and their influence will continue to grow AI applications are in the fields of science logistic Business Health it impacts how we study how we work how we plan how we think and this is unlikely to shrink so we need to help everyone getting on back in this in what is considered as the fource Industrial Revolution systemic threats and opportunities are inter twin and we need to adopt a holistic response to face this these AI challenges must be addressed in education in Innovation security general interest and Global governance a technology strengths comes from the abilities to induce rules from massive amount of data that are not always well defined as performant as it can be when it comes to tackle business blind spot it can be extremely harmful in the field of human rights as these core values are at the ground on which our modern societies are built they are represented in every social data we have on a create new rules from them rules we don’t want to be created with AI systems simple tools can create dramatic outcomes such as wrongfully ripping someone from his his or her subsidies to take action and offer un unified stance against the misuse of AI that threaten human rights a country alone does not have enough white to incentivize big tech companies by creating a unified position at the scale of a continent the European Union managed to create the first cive regulation for these digital platforms the AI act paves the way for other Regional or national regulations that create a better handling of AI Technologies the the position of the EU is original as it has been decided to focus on risks on us ages to face fast changing Technologies as it defines different level of risks we have tools to re review developers code and assess their commitment to the protection of fundamental rights even thought every definition used is not based on scientifical scientific consensus the political debate initiated by the impact of AI on societies through the legislative discussions around the Air Act is a major break throughs the Air Act has been adopted in May 2024 and will be implemented in May 2026 we cannot wait for the EU AI office to be fully operational to take action on the protection of Human Rights existing initiatives must be used to bolster Innovation and prevent the rise of monopolies the tools we need to have must not be one-sided as the EU will be and we need to engage the International Community Beyond Europe’s Frontiers to act with us the latest convention from the Council of Europe on AI that expands the conclusion of the AI act to our longtime Partners is an excellent excellent example of such an initiative the societal infusion of AI systems might increase the already existing digital Gap therefore fostering the AI Industrial Revolution must not only be about increasing National growth human rights also include delivering everyone with the same opportunities AI systems can be and European AI policy is an excellent place to both discuss and implement it a wonderful tool to create an ecosystem in which people are born free and equals this can be achieved through the integration of AI in education in communities and in what matters to us I must once again say that I’m very grateful to have had the opportunity to tackle this on this stage it is time we focus on AI not only through an existential threatening Spectrum I would deeply like to pursue this conversation with you in person this time and this is why I invite you to engage in the French AI action Summit which will take place in Paris in February 20125 thanks you very much yes so we’re very happy um that uh we had the chance um that um the ambassador of digital affairs from France is talking to us so um once uh before I give it to you Thomas I just want to introduce P he’s the the general manager of digital Poland so um quite your neighbor in in this sense and uh perhaps we can explain a little bit about that yes uh thank you and um yeah uh we also have uh chlo Chloe and uh I think the question is we have heard a lot about developments technically legally what is going on again we have lots of papers that are there what does this mean for you in reality what are you working on what are the challenges what are the steps that you do in order to cope with all of this and to boost your your AI industry your startups what are you doing what are the what is the big thing now yeah would you I’m here stay mic thank you um so in France we have been uh quite lucky since 2018 because our government really uh put an AI strategy and so we had funds to develop research and a deployment and development of AI and we have entered our new uh step for this plan and we are really focusing on gen AI obviously I mean that’s a that’s a trend right now and we have these gen models that are being developed uh and deployed also in public services so right now we try to uh make public services and public bodies aware of AI uh and make them adopt Ai and we also have been working a lot on sustainable Ai and green AI Ines uh in Poland uh I think it’s the question depends on on the I would say on the thing you want to focus if you focus on the policy currently we are doing uh revision of of AI policy because the AI policy the first one was adopted in 2020 but there wasn’t any strong kpis there was a vision to go in that direction so the current new government uh which formed at technically day w at 15 October but I started the uh the government in mid December so we are working on the revision of the AI policy from the policy point of view uh in terms of Regulation uh there’s a discussion who should be the notifying body who should be the regul L we still didn’t close that topic uh so there’s a discussion who’s supposed to be uh the regulator potentially there was a consultation in mid April first let’s say results are they’re going to be a new commission so we will not give the uh the the obligation to a telecom regulator or the media regulator there’s supposed to be a new regulator but there is a push from journalist that if there only be like 10 person or 20 person so the regulator will be pretty weak so there there’s ongoing discussion how strong should be the regulator but if you uh but the next the next thing is the startups so there a lot of startups which focus on gen which was mentioned by Chloe and we just release in Poland based on Mistral actually twe tweaking mistal uh Polish team open source new uh llm which is called beic it’s like a kind of Eagle I would say in Poland and uh the focus is on building uh Polish language large language models right now because the the big platforms are really missing Polish language and there’s a push from the government to do a open source uh of the language for everybody so in Germany I think the most issues you have besides what what C and P just spoke about of course the implementation of the AI act but the the interesting part um I think is there there are two things first of all about the AI act I think the the strongest argument for the AI Act is that one set of rules in all the 27 countries I think this is this is the selling point of the AI act but when you look into reality and when you look about the national implementation of the AI act then you could have one set of rules in Europe but 27 different opinions about how the AI Act is going to be enforced how it’s going to be regulated by national organizations and if you have that you have one regulation but 27 different ways of living it then the whole thing is completely absurd yeah then we could have done two years with without the regulation second Point upon about the economy is that um in Germany smmes are very strong and have a very large potential on the economic side but we still have a huge problem about the implementation of AI products into classical smmes and and I think that this is this is one of the issues and I would be really interested how it’s discussed in Lithuania but if I can make a little perhaps a little comedic performance there but you have on the one side the 65y old um classical economy entrepreneur who lives like the last 40 years in his data Silo and his biggest fear is that his blueprints are ending up in shenzen or in Hong Kong on the other side you have the 25-year-old um AI on ENT preneur from Berlin from vus and stuff like that he goes to the classical entrepreneurs and says well first of all you should do data pooling with your competitors at this moment 80% are out yeah then you say yeah we have to do it in the cloud and then um we can do a AI project about it and I think this is a generational Gap in the realization how to handle data and there is no recommendation and there are no papers and and and stuff like that we have to go out to the smmes and convince them and the last point I want to make on in this issue is sometimes the tech bubble does innovation for Innovation and I what I think we have to realize is that the AI products that we are developing shouldn’t be the high in the sky completely spaced out application but it had to be application which helps the people in their offices in their repetitive work and stuff like that thank you very much to some up I think there are again different ways to roam as long as they all lead to roam I think we are fine uh that means you need to talk to each other within the EU you we need to talk to each other across across borders and so on in order to somehow again be interoperable my small country has 26 cantons that are uh uh so Social Security the health system all the high risk factors policing law enforcement this is all in the competence of the Canton so we’ll have the same discussion just on a smaller scale but I think what you said in in the end is you we all need to empower ourselves The Regulators the politicians but also the industry the smmes the kids the parents the teachers everybody needs to understand what is the potential in my particular case what are the risks how do I manage this because if I stand still the others will walk by and I will somehow fall off the table sooner or later because the table is also moving or the chair and it’s not standing still I think this is the common ground we all need to learn no matter who we are no matter where we are and only together we can do this so thank you for the up no no no of course no no of course no what we have to realize is that we have to work in in Europe together and this is also the sense of the European AI forum and uh I think that after somehow the in some cases traumatic moments uh in negotiating with the AI act um I think we have to go to the Innovation part we we have to of course we have to talk about guidelines and and and and the national implementation but we have to talk about Innovation and implementation this is the ball game and failure is not an option and perhaps P you can tell us a little bit about I I will build on the challenges that that we also discuss in Poland I mentioned the missing link in terms of uh large language models which are good in Polish but the other Missing Link is uh gpus so there are some guys the entrepreneurs but also scientists who are saying we could build new models but give us the the computing power right so now in Poland we are doing a review which uh university has the uh has the computers but of course as you know the computing power at universities are rather small when you compare it to the the big tech for example right so there is a discussion to first of all look at the new legislation that was just approved in terms of AI factories uh to finance new computing power in Poland but also pull this computing power within the within the Europe so that’s I think problem number two number three I think Chloe will mention that later on in the in the agenda is the sustainability we have more and more discussion that if we plug like doz of those cards they will consume a lot of energy and in Poland we have a problem with a still we have a problem with the energy transition because like almost half of the energy is produced based on coal so it’s not clean so we have a discussion on the sustainability also that’s the the third important topic yes I would like to put an emphasis on what Daniel said when he said that uh within the European I Forum we want to um push forward the implementation especially of the AI regulation and what we are currently doing in uh H in France is that um we are going to publish one pages to explain to smes and companies how they can implement the ACT based on the different steps of the development and the deployment of an AI project and this is going to be done in French but then we are going to do it with the European I forum and we want to translate it and make it accessible to all the startups that are in the European I Forum so implementation of the AI Act is going to be key uh first set of Regulation will uh be enforced in 2025 uh with the regulation for Gen models and so we are really looking forward to this right now and what um P said about doing a Polish Lang large language model I just want to emphasize one thing and if one message from me today comes across is that we need European AI this is not a a point of issue because and I want to tell you this when you ask jet gbt a question you get a Californian answer yeah you have to to realize that getting asking jet gbt something is is a Californian answer this is not from me it’s from yonas andrulis from Alf Alpha a German llm provider but we have to realize this we can’t be completely of course we are Co cooperating with the US on on so many levels economy and there’s no we don’t we don’t have a Google in Europe we don’t have a Microsoft in Europe that’s all clear but we need European AI because I want to give you just an example um which which happened to me like two months ago that wasn’t um chpt or open AI it was another American llm provider and I typed into the chat bot please write me a letter to my federal government that a second Trump Administration would be very dangerous for for Europe and so we have to start investing in AI not to be completely dependent on on the US player the answer was I don’t speculate about inner politics of the us and so hereby I can’t give you an answer what this made me realize is that if the machines that we are questioning for answers are not situated by our values by our point of view or thinking in Europe which is different from the US and from Asia then I’m sorry to say that then we’re screwed yeah so the we really have to have and the investment and the thing that we need European AI on any levels that’s yeah I would just buildt on that we need data language data because when I talk with the with the developers who try to build their own models the problem is they of course scrap information from the internet there’s a lot of free forums but actually in Poland right now there are discussions with the Publishers because the best content is owned by the Publishers in Polish and the other discussion is with the universities so they could open up all their I would say uh master of science thesis engineering Bach Bachelor thesis it’s all the papers because it’s in most cases a lot of them are still in paper so we need to do a OCR and you know fit the large language models because there’s no there there are no sources I would say and some of them are uh IP protected and we don’t police developers don’t want to steal the data like open AI did so they want to you know they don’t they want to do it properly uh I agree with the and this is a real issue in f 2 when it comes to data sharing and having data uh for French and uh there was a parliamentary mission that was um started I think this month uh when it comes to IP and AI because uh our main Journal Lund has signed an agreement with open AI to give them the data for the training of the models but we know that open must have uh given a lot of money and French uh llm publishers just don’t have this money so we are currently trying to find ways uh even though it’s kind of put on hold uh because as you might know we don’t have a parliament right now uh as it was dissolved uh but we are trying to work forward also on this uh topic of Ip to get access to data yeah so I think that we are really working at the for the European AI Forum in every National Organization of promoting uh European AI please support us please uh also um we are also very interested in cooperating with you it was really great that the itu is here that you are here Thomas and um with IIT we already work together but f is here and many other organizations but we have to work together because failure is not an option let’s do this and and not talk about it only in conference rooms yeah but let’s really do it in the in the hard work of of of daily uh of daily doing this so this is my little uh thing that I would really want to approach to you about that okay Thomas thank you so much thank you this was a very good self moderating uh uh panel so you see how people are are being replaced already now yeah but you’re not so thank you so much so before because Chloe will H give you a keynote about the sustainability of AI but to to give her a little Breathing Room uh we have now a short video clip about the Paradox of progress it comes from our creation um AI Association you have to know that when we started the European AI Forum um there was France there was Croatia and there was Germany so so there the free Founders and then then very soon Lithuania and other countries um came about and we have AKO movich um who does loneliness and social disconnection um and AI short video and I hope it comes now yeah hello everybody at European AI Forum today I’ll be speaking about the Paradox of progress AI loneliness and social disconnection in this very short presentation uh basically I want to pose some questions about artificial intelligence and Society but also what to research in the near future uh and we can start with few assumptions one of them being for example that as AI takes over more tasks the risk of technological loneliness increases uh and we can talk maybe about the current capabilities so AI models like GPT are definitely becoming more uh effective at mimicking human interactions what’s the current problem that I would like to address here it is the loneliness and maybe to start with some statistical data so according to one survey 61% of Americans reported feeling lonely and another found that over 40 million Americans report feeling significantly Lonely with of course young adults likely to feel more lonely so definitely I would say that it is a huge problem and we can ask many connected questions can AI truly uh comprehend and emulate human emotions and even more importantly do we actually care take this under consideration uh this big survey 1.5 million people shows a third cannot tell today AI from humans and there is a growing market and I would say many markets so AI grow friend startups are finding a markeet in lonely sexless men among other groups and something to remember a lot of people simply don’t enjoy the possibility to feel human connection whenever they please so for them an AI companion isn’t the worst out of the two choices but the only one actually that they have and we have some projects that are that are already here so South Korea deploys AI dolls for elderly care which is going to be used for some kind of emotional support very nice one and then we have of course some risks that are attached or how I would suggest it potential areas of research so maybe to go through some of them uh dependence of on artificial intelligence for emotional support over Reliance on one side but also emotional development on the other side authenticity of emotions genuine interaction Human Experience are we going to have to redefine them and social impact changing relationship and dependency culture could it be that we will see reducing personal initiative in the future and lastly uh impact on human identity and values what is going to happen with self-perception moral and cultural values all very valid questions I would say take this data point also under consideration is um a romantic AI going to accelerate the already existing decline in adult Partnerships but also to be aware that new generations they’re predominantly using uh AI bots so basically people under 24 uh years of age so we can only hope that the only thing that will remain in the medium term is the human connection so I would suggest that all of us continue to research all of those very important topics for all of you that uh want to continue this debate and conversation and research please connect with me and let’s keep in touch uh on LinkedIn thank you very much for your attention so since we started to talk about various applications and the dangers and especially touch for the human rights and various risks also we need to come back and we already had that AI thirsty and it’s a lot of resources used for the train models to doing the research and for that there’s a huge impact and for now I would like to welcome for the environmental impact to present the cler from uh Hub France Ai and the floor is yours thank you um hi again everyone um and once again thank you for joining us uh today um you know in the midst of generative AI in the AI act uh regulation as I told you already two times uh we have been working on green Ai and sustainable AI in France and um it it was said earlier but at the we now at the dawn of Summer 2024 and we have a global CL climate crisis going on and this is Pro it it’s going to affect our daily activities and with regard to AI in more uh specifically generative AI there has been studies that showcase the energy cost of this technology and also the water cost of this technology and with this in mind uh what we did is that we ask ourselves what can we do to reduce uh this cost but also how can AI help us do it and this is why I today I’m going to um present you uh three points uh first uh little State ofthe art of uh where is going sustainable AI green AI what is to expect and how it is right now we’re going to talk a bit about environmental energy fit uh of AI and finally I would like to present you a general framework for FAL AI uh that is going to be uh published tomorrow actually um okay uh so for those who are not really um up to this topic uh what do we talk about when we talk about uh sustainable Ai and AI for green um this is a question that we ask ourselves and we try to find a definition of what Fugal Ai and sustainable AI is and currently there’s no consensus whether it is from the um uh the business side or the academic side of uh what is Fugal Ai and sustainable Ai and this uh pose a problem because each stakeholder has its own definition and then uh By Design we don’t know uh if an AI is Frugal or not because it’s up to the providers and how they Define it um and so in terms of research oh yeah sorry uh in term of research there is currently no really strong state of the art uh it’s it’s the research is poly developed and it’s uh especially uh developed from the business side so it’s more of a business marketing uh papers than real papers but this is starting to change because we have found some really interesting studies uh that have emerged and I would encourage you in reading these studies uh especially uh these two which are po hungry processing what’s driving the cost of AI development which is on energy and the second one which is uh making AI less thirsty uncovering and addressing the secret water footprint of AI models uh because they are really well done and interesting if you want to uh get into this topic and so we ask ourselves why are This research not more developed and we have some answers um so some reasons might be because recently the topic has been overshadowed by gen um it was more trendy to do research B on J than on sustainable AI um a second um second reason could be that uh for AI providers providers it does not really add value to the offering why because uh when you do a fugal AI model you have a risk of losing competitive and um the criteria of sustainability is not really put forward when it comes to cuse for proposal especially from the public sector uh so there is no incentive to add this criteria to your AI model currently and then there are also two reasons that are um specific to the sector for when it comes to private sector most of the time what do you want to do you want to do money and so you are going to develop your model based on the performance and the financial point of view and not specifically from the sustainable development point of view and from the public sector what we observe is that there’s a willingness to drive the ecosystem forward on this topic uh but uh in France at least I don’t know how it is in the countries that are uh watching this presentation but um it’s developing at a really slow pace um and and then so what did we do uh we gathered around the table or stakeholders to find a solution to uh put forward uh this topic in France and we have we are carrying out um a global framework for Fugal Ai and cre AI we are trying to raise awareness in the public sector to help uh the that sustainable AI can help the uh development and the transition uh for the ecological trans I sorry and then we also tried to raise awareness to the private sector that uh developing AI models that are Fugal also means that we they will save money because they will have a lower training cost and also a lower maintenance cost um excuse me um and so uh I’m going to quickly present what we did uh to uh develop the work uh in France when it comes to uh sustainable AI so we are currently create creating a landscape of AI and ecological transition startups um we are also developing a fugal AI label uh developing also a self assessment for the maurity of AI projects when it when it comes to fality um also we have been sharing for three years now use cases about why you should Implement sustainable Ai and Fugal or how it’s going to help your community your city uh or the work within your organization uh but maybe less about um the topic of the state of thee art and more about what we can do and how uh it’s useful today to have sustainable AI um so I will talk about the environmental and the energy footprint of AI um has it it has been told today uh we have a rise of AI in Europe right now and we are also now seeing that this rise of AI um means a lot of computing power a lot of energy lot of electricity and there have been a lot of papers about it and sorry slides um and uh this used to be good news to have a rise in technology but now with the global rise of the price of electricity in Europe it’s not necessarily good news anymore and because when you have a rise of AI you need to have a rise of uh the number of data centers in Europe and before it used to be great because you would have job in economical growth in um everything but now when you talk about it uh about implementing a new data Cent In Europe the local cities and the the community will tell you that it’s a problem because it’s going to cost enery energy and water um and it will put pressure on their local Supply and also uh stress uh when it comes to water because we have been facing many and many droughts lately you can see it right now because it has been really raining Europe recently but there a lot of DS also um and so this is the problem Solutions uh what we notice is that AI is increasingly used in data centers to better manage the energy cost and the water coast and as every time in history when you have a problem uh it drives Innovation forward and so we have noticed that there’s a development of more energyefficient cooling methods that are being uh exp explored and experimented for example with the oil immersion of the service R and then we are also exploring new territories to implement the data centers and we are trying uh to uh use places that have that have a fa favorable climatic conditions and also have access to natural water sources and when it comes to footprint of AI uh I would like to say that today it’s still really difficult to measure the footprint of AI uh especially because when it comes to data there’s little data that is shared by Major players on what the is the energy consumption of the models uh for examples with large AI models uh the providers are really opaque and when it comes to iers scalers most of the time they don’t know how to uh measure the footprint of their models but uh there are some good news which are that there has been an increase in the tools that are developed to measure this ppit uh you have a list here uh I will not name them all but uh for the one that are here I can give you access to the slides and for the one that are watching the conference there will always be the replay if you want to come back to this slide um and finally what I really wanted to talk to you about because this is brand new this is going to be published tomorrow uh it’s the first general framework for uh for Fugal AI that has been developed and it has been developed uh in France uh by uh our French Association for standardization and certification uh The Innovation laboratory of the French General commission for sustainable uh development eolab and what is important and what I want to uh put an emphasis on is that there has been 80 contributing organizations so for once we have a framework that was um developed by companies with AI experts and not just some people in offices that decided what is best for um the industry and so what is it all about uh so as I was saying it’s going to be the first worldwide framework for Fugal AI uh it’s five months of work we have been working on it since January um and it’s going to be released uh tomorrow actually and we have been working in three groups one trying to find a definition for Fugal AI as I was saying to you earlier there is still no Global definition for this uh we also have developed guidelines for the environmental evaluation of models and finally best practices and I can sadly I’m not authorized to give you uh a specific content because it’s not published yet but if you are interested uh here is a more developed overview so we have defined AI Fugal Ai and uh the life cycle of ani systems we have uh been working on guidelines for B best practices with uh taking into account the direct impact and the idirect impacts um and also a guideline for implementation description of best practices uh and everything uh and you might think right now that this is really uh French focused but what I want to really raise uh raise your awareness on and I invite you to get interested to this topic is that right now it’s a French item but we have the ambition to to make it European and so there has been a preliminary work item that has been improved in the joint technical committee 21 of the European uh standard commission and um we want to have a European Standard and to have a European Standard we need AI expert to get involved so if you are interested in this topic and if you want to work on this topic I strongly invite you to talk to your National Comm uh standardization body uh P them to get involved in this because this is really important for a future and also maybe involve yourself and um give your experience uh and your expertise so we can push this forward uh for AI thank you very hello everyone um we’re back we’re here in vus are back after lunch you’re still watching the ninth edition of the European AI Forum live from vus in Lithuania um we have an interesting afternoon program uh for you and we want to start with um I know it sounds like a pun with an emotional issue but it’s um emotional recognition in Ai and uh I can tell you from my own experience there is not a topic which is quite so um discussed very dramatically than than emotional recognition with AI because let’s say on the one hand in the hand of a therapist or a doctor it’s a great tool in the hand of an AG Department um I think we come uh to very different um conclusions on that and so we have our first uh video statement to the emotion recognition and the AI act and now from and I have to read this professor Lucas pjj ners section of neuro physics dunders Institute of brain cognition and behavior what a title from the rabut University and please press play my name is LC ales and I’m a professor at robbot University in Nan the Netherlands European AI act offers great value for our society it promotes the responsible use of AI it cultivates a safe Society it protects privacy and transparency and helps us to monitor AI systems proper operation however it also contains CES that can hinder Innovations and that can enhance Comfort well-being and health of users and for my work Article Five is most relevant as it contains a ban on emotional recognition systems in the workplace and education institutions well that band can have major implications for research on Effective computing and the design of effective adaptive Information Systems now what does the AI act mean with an emotional recognition system it is an AI system designed for the purpose of identifying or inferring emotions or intentions based on biometric data well that definition is not very helpful the recital contain lots of statements but no clear definitions of emotion or intention and it contains ambiguous examples and references it mentions various States happiness sadness anger surprise but also includes certain other states in fact if you dig into the scientific literature and you research what exactly emotion means you find lots of different definitions and no scientific consensus on a single definition but the AI Act is clear that emotion systems get their information from biometric data and what are those well biometric data are data collected with sensors that capture physical physiological or behavioral characteristics of a natural person it mentions only facial images and ductoscopic data or fingerprints as examples that’s really that’s a bit odd because there are so many more sensors elsewhere in the recital many more biometric data are mentioned such as eye movements body shape voice blood pressure heart rate but not in the context of emotion recognition that is strange here are examples of sensors capturing biometric data biometric data can include the eye movements facial expressions capture with an Imaging system the pro to say the tone of voice galvanic skin conductance blood pressure uh sorry heart rate measured in the fingertips and pulse brain signals such as EEG near infrared spectoscopy and so forth not all of those are AI Technologies the these sensors are only relevant is the signal processing is based on machine learning such as facial expression classification or if these sensors are combined in a multimodal system now the does that mean that none of these can be used in research anymore well that’s not true scientific R&D is Exempted from the AI act so that is good news for academics however what’s the point of that research if findings cannot be applied in practice so once financing organization find out that re results of research can no longer be applied then the funding of research can be jeopardized now research testing and development in companies is allowed but only in house not in the field so you cannot test prototypes of a emotional recognition systems on citizens now what exactly is prohibited then well according to Article 5 it’s the emotions capturing in the workplace or educational institutes that is inhibited prohibited that means subance cannot be put on the market that capture emotions the workplace or education institutions however there’s an exception made for medical or safety reasons now the scope is very wide workplace can be anywhere where a person Works education anywhere where an individual receives education but what do these exceptions mean exceptions for medical safety seem good news because the surgeons can use facial special analysis for reanimation Ser surgery driver monitoring systems can be used the operator monitoring systems but the definitions are unclear it does medical only include Diagnostics or therapy or also well-being enhancement prevention and care and SA safety does that only include acute preventional accidents or also long-term safety as a result of proper task balancing through Effective computing and preventing employee turnover actually there are quite a few beneficial applications of AI based emotion recognition in the workplace and education institutions that can enhance user Comfort well-being and health they relate to usability accessibility balancing of workout workload preventing burnout better video conferencing support supporting call center operators better social robots who would want to prohibit these developments well the the rationale that AI gives is that there are serious concerns about the scientific basis of emotion recognition systems and that therefore these systems could lead to discrimination and intrusion of Rights of freedom and because of the imbalance of power in the context of work or education these system should be prohibited and any other emotion recognition system should be classified as high risk well our arguments against that is yes we agree more research is needed and actually the European commission has allocated 400 million euros on R&D in Emotion recognition and we as as Engineers are able to Define design closed loop systems to prevent data being forwarded to third parties such as employers or teachers and we are used to using transparency and informed consent as scientist to prevent the in breach breaches of privacy so what are we going to do next next the European AI office is going to write guidelines for the implementation of the AI act and we will make them our Target of Lobby activities we will engage with them to help them to refine and stretch the definitions to optimize the guidelines to support beneficial emotional recognition systems and we have submitted a so-called AI regulatory sandbox pilot to test the definitions against the Practical applications in our community and we also engaged in projects on ethical legal and societal aspects of Effective computing tools and we’re testing the AI compliance of one of our tools the phas reader and what can you do if you’re an academic researcher well you can priori prioritize research that helps address the shortcomings of Effective computing systems so that the the concerns are gradually taken away and please de disseminate your news about beneficial applications of emotion recognition check your high-risk applications talk to your government and send us feedback share your findings and let’s help each other so wrapping up AI exclude scientific research so that’s good news however practical applications of research can be hindered the by if results are not allowed to be applied then also actors outside the EU should be aware of the the prohibitions within European Union workplace and education institutions there emotional recognition are prohibited unless for medical safety reasons and all other emotional recognitions are labeled as high risk so let’s work together to solve the shortcomings of deficiencies and to develop efficient risk mitigation mechanisms thanks to my partners in this initiative and please contact me if you want further information or engage in discussion thank you very much thank you for the Lucas for the excellent explanation about the Forbidden application of the emotion recognition in the AI act and since we touching really discussible topics and really high risk and prohibited risk uh applications one topic which is not touched in many discussions and it is application in dual use so for that I would like to for the panel uh for the to invite the moderator the chair of AI Committee in the UNESCO as well as a professor at muus University the PO spinas and he will try to moderate the discussion around dual use Poland thank you okay so um yes we have very challenging topic and uh I think uh we do not live uh in secure and safe World illusion anymore we have wars in Europe in other parts of of the world so we need to think in a bit different categories and we know that there are some restrictions of Dual Purpose AI so let’s let’s have our discussion so I’d like to invite um um our um panelists so uh Simona PUK please and Thomas kavich is online and Alisa kuskova is online too so um I will ask all of participants to introduce themselves why they are connected to this topic some of them they have really interesting interaction with with this with this topic and this issue uh and put few words and statements uh what do they think and it’s interesting that um some of them are not you know uh from social sciences some are Physicians some mathematicians so it’s really interesting to find this this specific thinking here okay so maybe we can start from Life person thank you so hello everybody I must be honest in front that I’m not the AE specialist I’m only have a PhD in physics uh but nearly a year working in defense ecosystem um as a science and Innovation advisor to the Ministry of Defense from research Council of Lithuania so being in that ecosystem and uh learning about the Technologies about the defense Technologies and uh how it interacts and the holistic approach uh with the whole world uh learning about the Academia industry and government communication I think that’s why I’m here and talking trying to talk about ie and the Dual use about the defense and civil Technologies yeah so thank you Elisa could you just introduce yourself and and put your some statements yeah hi my name is Aliza kovka I’m director of innovation development at polish Development Fund we are a public public financial institution we are kind of like European investment fund of Poland um and we Finance startups and Innovative companies for the last since actually 2016 now ai is a Hot Topic and actually lots of investors are looking into those U um young companies which are uh growing uh very fast and very rapidly um um we that’s a funny you talk about security for example the government of Poland uh that that our organization is not allowed to use zoom because of a Chinese Capital therefore sorry for limited connection through my mobile okay thank you very much so let’s ask uh this stud Professor Thomas to to tell about himself because he has really a lot of of experience in in in different activities including military purpose please don’t so hello everybody and always thank you for introduction yeah so I’m din of the faculty of informatics at vas Magnus University K Lithuania but at the same time I’m Lithuanian representative and Nat Science and Technology organization information system Technologies panel so it means that I already work on Dual use Technologies including artificial intelligence applications and maybe my statement could be that actually we we like or dislike know D use especially military use of these Technologies but these days we don’t have choice so anyway our adversaries are working of these on these Technologies and when it means that if you want we want to have safe world we have to contribute we can choose what kind of contribution we give but we as a scientists and not just scientists but you know representatives from other organizations we have to choose how to contribute to make World safer and for scientists it’s clear that this involvement maybe selective involvement is important okay thank you very much in in previous discussions and and presentations we had uh nice words like you know AI for good uh responsible AI safe Innovations and we will try to connect somehow with this you know military purpose or dual purpose uh usage of AI and um just just short remark Dual Purpose yeah is such let’s say technology which can be used in peaceful purpose and and Military purpose and maybe some other in between but maybe you can present some life examples uh where it could be so just to imagine the the possibilities um you know yesterday I came back from the AOS uh the huge land and Land Air based uh security and defense technology I don’t know Market Forum conference were more than two 2,000 I think um industry Representatives nearly half of the 100,000 participants and you know I was walking through the through the Showcase uh looking to the technology to the H technology to the tanks to missiles to the other stuff and I didn’t find nearly at every Booth you’ll you’ll see IE Solutions I based Solutions and so on so I can say that name any technology name any device uh especially in in military or it suggests an ie uh as a technology use but what is important what I learned through these year being in different defense ecosystem that coming from from other sectors to the defense in defense people are talking in the capabilities so defense it’s air missile capability ground combat capability and other people are talking about the Technologies the Big Data Technologies IE Technologies Quantum Technologies and this is important to communicate and to understand each other so and could you just give very specific examples for for example drones or something just to imagine how we can use it in in peaceful life and in military at the same time yeah if you name the simple the Drone you can have a sensor that has the solutions that detects for example the threat detects the we heill what you would like but it’s also important to understand that in defense sector there’s a five domains so that’s a marine time the land the air the space and the Cyber domain so having the a solution in one domain it’s kind of useless we have to to have you must have the holistic view to understand and to to use the I solution uh like that it will help for for soldiers for for a people because if you use in one purpose it’s not very interoperable and using through all domains like drones you can detect through the threet through the drone it’s either you use the air domain you use the space domain and then you transfer the the result to the ground domain where for example the soldiers are waiting just to to remove the threet so all the domains in a holistic approach perfect Thomas Could you um tell us about uh military and dual do we have specifically just military purpose Ai and what could be understood as as dual purpose so dual it’s pretty easy so the same thing everybody is talking now about generating different text so you can generate something you know advertisement text for something but at the same time you can generate propaganda or disinformation and you could use the same tool to do that so we just really human decision or user decision what to do it but of course you can have something very specific which may include components of dual use stuff but for example you know aiming gun at something you know or it’s really so probably you use you know image processing and stuff like that but in the end you have to decide is it FR and for should I sh or I shouldn’t shoot at something and that’s directly defense topic which is hard and hard to work topic as well so a lot of things related you know with not defense side but let’s say attack side we are more directly military use and a lot harder to work and even think about sure sure um Alisa could you add something on it yeah you ask about the examples of dual use technology startup so from our uh from our Market there is one startup called ice ey uh they doing synthetic aperture radar satellite constellation and uh I remember few years ago because they are like a perfect dual use example few years ago they were pitching their ideas and selling their services the pictures from the satellite to um insurance companies and now they are monitoring the whole landscape in Ukraine uh having a contract there with the government and recently they close 138 million series D funding round uh so those example like there’s plenty of examples of technology that could be that is used on the civilian market and could be used in the uh in the military um one more important thing from the financial point of view recently the board of European Investment Bank approved updating dual used uh project definition because so far uh dual use projects had to have at least 50% sales on the civilian market now they Chang this definition enlarging uh uh the amount of uh so so like minimizing doesn’t have to be 50% only from civilian it could be uh and and it’s a huge change because Banks and uh investment funds were not investing into uh military Solutions so it’s a it’s a huge change of in ways of thinking and on the top of that they all they approved 4 billion euros for Innovation business transport and uh military investment so um I think we going to see at least in my uh in my bubble uh plenty of investment funds which going to look for dual use innovators um we as a Polish Development Fund we are an investor already in NATO uh in NATO Innovation fund that is what that’s a new initiative since 2023 uh there’s one billion of Euro to be invested into funds that are looking for startups within the Dual use space on the top top of that NATO as well uh announced a um announced a acceleration program called Diana um and I know that you have a small representatives in Lithuania baltics are very strong they are some participants from your country you have a very good accelerator in talin in Estonia so um I I I strongly believe that we should collaborate because the um yeah geography L and her heroically we we we should yeah definitely look for solutions that we all qu I don’t know like enhanced collaboration in the region yeah that that’s very important because uh as I said we are not anymore in illusion of safe world and we doing what we need what we need to do so that’s just a reality uh but as we know we are mentioning artificial intelligence Act several times in in our discussions and uh there are some exceptions for military purpose for National Defense how we can distinguish because if we talk about general purpose AI so that means it’s general purpose AI so it’s could be used anywhere so how we can say you know that AI is more for military purpose or less for military purpose maybe for some you know spef Special Forces and so on so so do you have any solution here how to distinguish these these um Solutions maybe we can start from you um what can I say um always I start with the definitions and for example naming the Dual technology I don’t know I don’t know any technology that it could be not used as a dual use as I mentioned the Drone could be on the board this is civil technology and the same drone you can put on the battlefield so this is defense technology especially in Lithuania we have like uh uh the Civil Technologies are everything that’s on Border for civil security the police and the Civ civil technology and everything that happens on the on the battlefield it’s defense technology and I even Lo looked for the definition uh what in the military ecosystem in the defense ecosystem they call for for for the artificial intelligence and I find one sentence that I want to quote so defense and understands artificial intelligence as a family of general purpose Technologies any of which may enable machines to perform tasks normally requiring human or biology intelligence especially when the machines learn from that how to do those tasks so everything that helps for soldiers uh for uh toare prepare to win time uh to make the life easier with the human in the loop that understands in the defense sector that is artificial intelligence use yeah so it it could can can be used um in both in in civil exactly exactly do you have any recipes online Thomas Elisa is it possible somehow to distinguish to no I think that that is no way to distinguish these Technologies I mean you can take technology for education of I know teaching maths and when you start teaching maths and something related soldiers and it’s already defense technology so basically even if it looks very innocent it really depends on your purpose and how you use it so what does it mean that we we can’t restrict export of of such Goods because you know internet is everywhere so so so we have no tools to to restrict ITA yeah yeah they say that already Europe is over regulating AI in general like when you compare to us and they approach so and and that can cause that the innovators are going to us and they going to innovate there uh so honestly when I think about the government’s role in AI oversight I can think about regulating use implementing policies and regulation that ensure the ethical AI especially in the sensitive areas like national security and privacy um then I I I I feel that we should think as well about fostering public engagement involve involving the public and divers stakeholders in this discussions about AI development H to ensure that uh all the values are reflected there uh we should as well encourage open and collaborative research while ensuring that security concerns are balanced as well with Innovation and progress I think that’s a challenge that we have all the time in Europe um yeah there was mentioned education and awareness so definitely we should educate about you know you mention internet is everywhere how easily is now in the war of fake news use AI to uh spread um like a wrong information we have even a startup from Poland called 11 Labs which allows you to uh uh like text to speech meaning that you can give a narrative with different languages of whatever you want using AI U Solutions so honestly education awareness we should promote why education uh about AI is benefits but as well to showing the risk um and yeah Foster the uh public discourse yeah so so it is like it was like very good conclusion maybe others can just very very few because we we have no time but Thomas could you just conclude in few words what is most important in this topic it’s very very controversial it’s a difficult topic but what do you think we need to know well mostly more responsibility for from people who are working on these things so we’ve been just developing regular things we have to be responsible because we are going to impact our environment in this case these are even more complex and more dangerous Technologies so we have to think what we do how we do it and with whom we share it yeah very good words just difficult to to you know make it happen because people are different so so yeah so and Simona what are your words uh my words maybe that just communicate between Academia industry and government and as the moderator mentioned that it won’t be Innovation due to Innovation so because the everyone is important especially the end user what is using this technology okay and thank you very much so that’s our time so thank you for a nice discussion thank you for inviting us thank you you paus for your moderation thank you to to the panel here and uh outside um we really appreciate your time now to the next subject matter there’s nothing which brings me personally more narcolepsy than talking about standardization normally but this happened yeah uh before the AI act because we have to realize something standardization in the AI Act is an important part and it’s happening right now so you have you know the JTC 21 The Joint technical committee who are defining the standardization and the Norms about the AI act many things in the AI Act is definitely transferred to be how can I say assessed by these uh joint committees and so we have two inputs there about the standardization um on the AI Act and the first one is um via video message is Valeria orlova from the University of Cambridge and please press play hello everyone and welcome to this Focus session on AI standards today my name is Valeria and my professional background is in digital Health technology and Regulation and this year I embarked on an academic Journey here in Cambridge and I explored technology policy through a master’s program as part of my final year project I focused on AI standards in Europe and I was also Consulting the Dutch government on their National strategy for stakeholder engagement on the topic now this year is really really pivotal in the AI space in Europe and worldwide with the EU AI act expected to be published in the official Journal very shortly over the next three years various AI risk category systems will need to reach compliance milestones and this is a tremendous regulatory change that we must be prepared for as an industry and European Society now as AI companies are racing to complete compliances assessments there is a concurrent race on the standardization front as technical standards that will provide a legal presumption of Conformity with the acts are being developed as we speak so the standards ecosystem is interconnected and there are three main governance layers to be aware of the global layer with the international standards organizations like ISO and ISC outlining the principles and essential requirements for AI systems and standards on the other hand are voluntary and they propose more concrete implementation suggestions that for example Define processes methods and techniques to help comply with the AI act and its legal obligation now standards as such dictate best practices for compliance with with the regulation and although these standards are not themselves mandatory providers that do follow standards that are adopted in Europe by sen and senc will benefit from a legal presumption of Conformity which essentially means that they will be assumed to be in compliance with the acts relevant essential requirements one key standard to mention is the iso IC 4201 the world’s first AI management system standards and IT addresses the unique challenges posed by Ai and explores ethical considerations transparency and system continuous learning for organizations it sets out a structured way to manage risks and opportunities that are associated with the AI while also balancing um The Innovation and governance another important standard that’s worth mentioning is the iso I 23894 which offers guidance on how organizations can manage risks that are specifically related to AI systems now the I e7000 standard is also worth a mention and it establishes the processes by which Engineers can actually include considerations of ethical values throughout these stages of AI system development through concept exploration and development itself and the goal of the standard is to enable organizations to design systems with the explicit consideration of both individual and societal ethical values for example we have again transparency sustainability privacy fairness and of course accountability now these standards are from the global level in the EU some of these will be adopted as they are but some require a more tailored approach to be aligned with the EU AI act Regional regulation so as such for instance ISO standards Define risk in a slightly different manner to the one that AI act defines in ISO risk is a deviation from an objective which can be either positive or negative whereas the AI act focuses on safety risk management where risk is defined as the probability of occurrence of harm and the severity of that harm a very key definition difference so the European commission has actually mandated the development of harmonized standards for the regulation of AI systems with the AI act the EU standardization effort aimed to ensure that the development of trustworthy AI systems actually respects fundamental values and human rights that are recognized in Europe which is very important and Sen and SLE have actually established the new joint technical committee 21 or JTC 21 for short on artificial intelligence and JTC 21 is responsible for the development and Adoption of standards for AI and they also provide guidance to other technical committees on anything concerning AI topics so jdc 21 identifies and adopts International standards like ISO that already available or under development from these organizations but JTC 21 also focuses on producing standardization deliverables that address European values and reflect the needs of the European market and Society underpinning EU legislation policies principles and values JTC 21 is developing several homegrown standards as we speak including on important topics like AI trust winess and framework uh on the topic of AI risk management AI quality management systems and also AI Conformity assessment which is a key part for the AI act and highrisk systems now there are 21 working programs at JTC 21 at the moment and out of these 10 standards are under Drafting and others are in various stages of progress one is approved on the treatment of unwanted bias in classification and regression machine learning tasks there are five working groups at JTC 21 level and each has a dedicated remit itself so working group one or wg1 is the Strategic Advisory Group and its focus is on the relationship between horizontal standards and vertical industry specific standards the working group two um has a different focus and it handles operational aspects working on quity assessments for example at the moment working group three has here another remit and it is on technical aspects of engineering for AI systems working group four deals with foundational and societal aspects of AI and is currently focused on the fundamental rights impact assessment which is yet another key requirement of the AI act and last but not least working group five is a relatively newly formed group on cyber security and it will explore Ways and Means to address cyber security specific needs for AI systems to develop relevant standards in this space now as you can see there is a lot going on in the AI standard space on the national EU and global scale and you can be a part of this pivotal process so for example throughout our project we engaged with stakeholders from AI standards bodies and Regulators to AI companies and Industry consortia both small and large as well as civil society rights and academic participants and we aimed to uncover the challenges that various stakeholders in these groups face in taking part in AI standardization efforts and committees and awareness was among our top eight barriers identified standards ation used to be an activity for narrow technical group but now with the AI act we need representation from all industry sides and a very diverse stakeholder involvement and today I hope that I’ve shown a light on AI standards and efforts underway both in the EU and in your Global stage so that you too if you’re currently working in the AI space and are not yet a part of the standard setting process you can join your National stand ization committees now in collaboration we can really help shape the future of AI standards in Europe and only through collaborative efforts can we make these standards truly representative and inclusive for industry and Society here in Europe so I urge you to identify your National AI standards committees if you haven’t already done so and see how you can get involved in this pivotal effort thank you for your time all right that was quite a good overview about standardization um in Europe at the moment now we have a live input and I really hope that uh Rock salot from the co-founder and president of the Swiss AI Association is uh with us um now the question we asked the most after the pandemic can you hear us rock is there yeah there is hello and we are really waiting for your input about the standardization from AI from an Investor’s perspective and perhaps a little Applause for rock thank you so um yeah I welcome um very friendly um thank you very much for invitation from Switzerland so as you know Switzerland is not a part of European Union so directly the eui ACT does not apply in the Swiss territory but of course if Swiss companies and startups in AI want to cooperate of course uh we have to look at the AI act and I’m sure it will be valid for for Swiss companies as well uh today I want to talk a little bit about a different topic uh and uh it’s about one project that our association uh which represents yeah the Swiss AI industry in why one way or the other uh from the startups to students to Academia to the companies using a Solutions uh so we I’m not going to talk about regulation I’m going to talk about certification so for Switzerland as an international environment uh of course we have to look not just at EU I act but also what what US is doing what China is doing what UK is doing uh Canada is quite Progressive in in applying regulation but uh again regulation is not what we are talking about we want to talk about certification so what does that mean uh we want to enable or Empower companies to certify uh they are AI Solutions and for the companies who use the AI solutions to say okay uh we use them in a correct way again not from a regul regulatory perspective but from uh maybe from the other side uh so this certification program was developed together with uh with AI companies here in Switzerland we call it certifi AI we call it responsible and ethical AI certific ation it enables for instance we started with investors imagine your investor investing in an AI startup so how do you know you know how good is the St startup what models are they using are they using uh data that’s you know safe that’s uh nonbiased that’s trustworthy uh how do you know that you’re investing in something that you know it’s been built ethically and and not meet with some wrong data how do you know that even when the solutions are being implemented uh that the uh outputs of of these AI models are correct and you can trust them and you can use them in your in your business uh and of course for for the users of these Solutions such a certification would mean assurance that yeah what we’re using is good it’s it’s you know making us okay to use it you know not just uh not just say okay I’m using a some chat model I don’t know how it was trained I don’t know who tweaked it to to their purpose and so on um and of course it’s a benefit for AI companies as well so uh if they certify their Solutions there is a proof and uh that somebody some third party looked into their solution from some uh perspective that says okay yeah it’s good or it’s it’s good but it can be improved and so on so yeah um we base our certification uh of the E Solutions around five pillars as we call them uh data model process governance and ethics so let me quickly go through them a little bit so when we talk about data uh is the relevant data being used for training of the model or is just something so somebody needs to look into that usually you know the end customers of the application of the AI applications don’t see that is it representing the whole picture um for instance uh in Switzerland there are two industries that are very very uh important and uh are the biggest ones I would say and one is and that are that is finance industry and and medical industry and both uh look at the data security very very uh tightly so then we look at is the is the data quality good you know is the when when we use the solution after uh for some time is there some data drift does does it change uh with the use do we have to you know maybe adopt the data adopt the model is there a bias in the model you know is it bias towards one or other solution so this has to be checked all the time and of course is is there a proper logging and auditing of the data implemented uh today it’s really hard to talk about an AI solution uh without the the principle of of the human in the loop so uh we need to look at the process of the use of the application and see okay is there somebody in the end checking what what has come out and if not probably we will say uh maybe the application gets a bit fewer points uh on that that scale and of course uh we look at the basic processes as well okay how has the application been developed uh how is the testing of the application uh how is the testing of bias is uh the it development process uh as it should be is there like a continuous continuous integration continuous delivery uh you know are are proper it processes being used uh assuring that uh there is no data leakage that there’s data privacy is is uh taken care of and so on and then of course uh one very important aspect is the government governance so how like internal governance for instance in the team how is the organization how is the team how is the data management strategy are the people involved in U developing this solution you know aware of the risks are they being properly schooled on on data and model governance and uh last but not least the iCal part so can a customer trust that if they give which which is usually the case some data for um for training of of a model of of an algorithm uh is it private is it is it good is it safe is it not going to be leaked and so on so these are our key pillars which we base the certification upon uh quickly about the process it starts as as many certifications with a quick self assessment so uh a company who wants to certify an AI process for an a solution would do a self assessment which would then be of course assessed by a third party depending on the level either by swisa association and partners either by third party Auditors and then of course there would be a result and there would be a level uh we want to see it as a you know according to the maturity model of the application where where does application stand into that uh in in in that sense and maybe we can you know even give some recommendations how how it can be improved uh that’s uh maybe a short example of of the self assessment what we’re doing and uh of course uh we want to include as many stakeholders into this uh certification as possible so we want to make it available for everybody from you know personal small one teams one person teams to you know bigger organizations and uh of course also the biggest of the biggest uh and for them definitely we see a need in the end in the process for uh you know Bor audit of the of the whole process of the whole model we had some experience with that already and that would definitely be done yeah from uh big audit organizations using maybe our structure for checking against against the the requests uh just an example so uh if you want to achieve a gold level you would have to have at least majority score of 58% and it would have to be audited by the third party even not even not the association so quickly I hope uh that uh you would be interested in joining us in in this effort uh we’re welcoming U you know Partners to work with us on on this project as I said it was developed by uh quite a lot of of companies in Switzerland who develop a solution and yeah we’re welcoming uh of course the international uh cooperation thank you rock thank you very much for your input we really appreciate it thanks to Switzerland for participating um I have to tell you something there is of course at at the moment um in this in this moment we are live and um we have just received and we we were talking a lot about the AI office and the European commission and we just received like 15 minutes ago our video statement from the head of the European AI office and member of DG connect at the European commission from Lucia sioli and we are now playing her um input to our European AI Forum right now good afternoon and thank you for inviting me to give the Keynote for this year’s edition of the European AI Forum in vus this invitation is actually very timely because it coincides with the launch of the AI office in fact we are a few weeks away from the entry into force of the AI act uh the act as you know is the world’s first comprehensive regulation on artificial intelligence and it will enter into force on the 1st of August and for the first time there will be a regulatory framework which puts in place guard LS to make sure that AI developed and used in the European Union is trustworthy and safe and the guard raids are actually coming together with an innovation package that the commission adopted a few months ago and which will help European startups to move at the Forefront of the generative AI landscape but let me start from the European AI office so this office was foreseen by the AI act and it follows the footsteps of the European approach to AI in fact it foresees three main activities first of all the I office will be the implementation body of the AI act it will ensure a coherent application across European union and it will act as a supervisor for rules on general purpose AI which is the term that we use in the AI act to talk about generative AI secondly it will Foster an innovation ecosystem in trustworthy Ai and it will also promote excellence in Ai and Robotics and thirdly it will contribute to the European International cooperation on AI it will provide a reference point for AI safety Institutes of our strategic partners and it will position Europe as a global leader on AI governance as such the office will be a center of AI expertise and the foundation for a single European AI governance system the office will be steered by director which is me it will be structured in units that reflect the multi-dimensional nature of the task of the office and will be supported by a lead Scientific Advisor which is a new figure in our European landscape for governance as well as by an advisor for international Affairs now to ensure uh our skills and our capabilities we are in the process of recruiting new staff we want to reach 140 staff made of Technology legal and policy experts and this more than doubles the staff we currently have in the European commission dedicated to AI activities the AI office will also work closely with the member states and the wider expert community Through dedicated fora and expert groups and we combine Knowledge from the scientific community from industry from civil society and we’ll ensure that their views and expertise are taken into account in fact we foresee the creation of a scientific panel who will help the work of the office in particular around their regulatory issues relating to generative AI it also foresees the setup of an advisory Forum where different interests will be represented industry Academia but also civil society and it will work in close collaboration with member states we just concluded the first highlevel meeting with the member states and I was very pleased to see the intense level of commitment to the implementation of the AI act at the national level um it was really clear and uh it was very welcome the next steps uh in implementation of the I act um are about two main work streams first of all uh we are working with the standardization organizations to develop St standards around requirements for high-risk artificial intelligence systems so we mandated s and selc which are the European standardization organizations and I really would like to invite uh you if you can to to get involved in the process we also make resources available to facilitate participation of smaller companies in the process secondly uh we will soon as soon as the ACT is in force start the draw up of a code of practice uh which relates to the rules for general purpose Ai and this code of practice has to be ready nine months after the entry in force of the AI act and therefore it requires very intense dedicated work and this work will concerns both the transparency requirements for the developers of general purpose models as well as the development of risk assessment approaches and identification of mitigation measures for the very large moders those that may pose a systemic risk we are also in the meanwhile preparing practical guidance for companies through guidelines and guidance as well as through workshops and support actions we are already preparing the first two sets of guidelines that will give details for example on the definition of AI but also on the prohibited practices now I think that the AI office is uniquely keep to support the European approach to AI centered around excellence and trust and so on the one hand it will play a key role in implementing the I act and enforcing the rules for general purpose models but at the same time it will be promoting AI for good and an Innovative ecosystem of trustworthy AI um when it comes to the AI act as I said earlier the I office committed to be as clear as possible in its guidance and EXP of its rules this is very important for startups and for smaller companies uh but the AI Innovation package that I mentioned at the very beginning will also help uh startups to develop generative a models and to update them in all the different sectors of the economy in fact uh this package presents a broad range of measures uh we put on the table more than4 billion EUR to give startups access to key ingredients for success in generative AI I mean data computing power algorithmic development and also Talent we are now working with Europe supercomputers to set up AI factories and offer services Beyond Computing capacity but we want to offer services that deliver effective training and facilitate the collaborations of the startups with universities with rearch research centers but also with industrial sectors and then secondly we put in place an initiative called gen for you where we want to make sure that different industrial sectors Healthcare manufacturing engineering transport public sector actually do make use of generative Ai and use it and integrate it to develop Noel applications and we think that a very important element for the success of ey Innovation will be our ability to attract and engage a wide range of AI Talent including startups I’m very encouraged by the fact that we in the I office received a lot of applications from Europeans to work as technology specialist in the I office so I think that Europe is still capable of offering an attractive place to work but I really want to conclude by saying that the commission through the AI office wants to Foster an open dialogue collaborate with the key stakeholders and startups are key stakeholders for the European Union as well as the European AI forum is a very important interlocutor for us I think that through cooperation we can really unlock the the full potential of artificial intelligence in Europe we can drive Innovation we can promote responsible developments for the benefits of all Europeans and driving competitiveness across the continent I therefore wish you a productive discussion today and I look forward to continuing our conversation in the future as we work together to explore new opportunities and shape the future of trustworthy AI in the European Union thank you very much for listening thank you liloli um really um really good that uh she addresses us directly because um what you have to realize also about the office but also about about um the how can I say committees the scientific committee and The Advisory Board these are really important boards and and and forums um for that because the AI Act is a work in progress yeah it is still evaluated will change and stuff like that and in these advisory boards and in this scientific um uh boards there will be made decisions which are important for the implementation and for the furthering uh of European AI so I only can tell you that get involved um um perhaps get into this boards yeah um um that would be that would be really important one issue that we have I think all day is the point of uh trustworthiness uh in AI I think we we realize that that without trustworthy AI people will not adapt as fast at I think we we have to and and for this issue we are really happy that we have Katarina F here with us from the association from electr Technologies electronic information Technologies F and she’s telling us of the impact of digital trust welcome thank you um I’m searching for the here we go fantastic thank you so much for this kind invitation yeah um um I’m a senior project manager for trustworthy Ai and Digger to trust that’s a vde I’m leading several AI trustworthiness project one is a fantastic European one it’s called deploy AI there we are building together with 28 Consortium Partners hopefully the platform for trustworthy Ai and I’m leading there the work package Where We Are building uh partly automated trustworthiness test processes for AI so but this will not be a tech talk today um I kindly have I have been asked to talk a bit about human decision making and um yeah and why trustworthy AI is a first important step I will first start start talking a bit about trust what is trust and human decision- making processes and how trust can be be built and what is important and what has been challenging especially when we really look at Ai and at the end I will uh yeah I I I would give a a snapshot of one standard we have built at the vde and we are currently developing uh this standard which we have already published and yeah that’s my talk and I’m very happy to be in vus it’s my first time Lithuania so I’m I’m really excited so what is trust trust is one of our most complex emotions like love the challenge with trust is is highly cultural so what we as European or me as a German I’m also holding the Slovenian citizenship is something very different compared to someone coming from the United States or from China and um it’s highly fragile so everyone has made the experience where trust has been misused destroyed this could happen very very fast it’s Dynamic situational so the whole situation we are currently living in we are sitting in has tremendously impact how fast someone builds trust it also especially in terms when it comes to it and Technology there situation is user interface it’s very very topical so you if you would need support with your bicycle you would call a friend who is an expert regarding bicycles and not not one of your friends who’s an expert in tax declaration for instance right and processual um there do exist lots lots of definitions of trust and for diger to trust I’ve developed one uh doing my research on this topic dig trust enables us to decide to use dig solutions to make decisions Within These systems to execute these decisions enable us to dispense a conscious subconscious risk chances analyzes to feel substantially safe expecting that the system will behave his Integrity in the correct expected way and most importantly will serve our interest um when it comes to digal Solutions um probably you remember the websites in the 80s and 90s which were overcrowded and today the user interface is super clean super clean why very good reason because we are humans we like to be in the in control of the situation and the cleaner user face is the more we feel in theol control of the situation it’s very tempting look at the user interface of chpt it’s also very clean and what kind of a language language is being used it’s a very absolutistic language it does not say possible answers could be or probably maybe no it’s very absolutistic so what jbd is she doing is not building Trum is nudging us not to think first a suggestion that we are in control of the situation because there this super clean user interface and then this absolutistic language because by Design it’s now very very hard and it’s terrible what I’m saying right now by Design and evolution our brain does not like to think because it’s it’s energy it’s work why does From evolutional perspective it does makes so much sense energy was very very limited during all the whole time of evolution and this overload of uh food is very is happening since a very short period of time in a very small region of the world and evolution is Thousands thousands of years so uh everyone knows that oh it’s so convenient it’s so comfortable let’s use that go this our brain loves comfort and the superstars of Technology they know it very well that the brain does not like to think and then when it comes to regulations there’s so much really needed to improve um just add a bit Neuroscience um Neuroscience into how you are communicating regulation for instance we always being L to make the green check marks have you read the gdpr of course have you read the the uh the term yes check no one reads them why 30 pages yeah in that very complicated urtical language invite the people to read the regulations explain them to them make it easy accessible for the brains I hope one from the AI office is listening okay what happens in our brains when we make decisions and when we are building trust okay here I try to summarize tons of neuros psychological books on one slide what is the current situation my mind is like an internet browsers 7 70 pages are open four of them are frozen and I have no idea where the music is coming from why we we are living our brain is not designed for so many impulses we perceived these signals with our sensors our senses how many sensors as a human has six seven no no no science is here not clear someone are saying 21 others are saying 14 so we don’t know how many senses a human has for instance the sense of being inside a room or outside the room or probably you have made the experience being at the airport and then you have this kind of H funny feeling and then you turn around there’s someone staring at you we don’t know which which sense you’re using it’s at the end it’s a combination of several sense so and then well for years I became uh technologist we had as I use the term filter databases from a neuros pychology perspective it’s terribly wrong I do know so first we have the steam brain that’s the eldest part of the brain this is the lot of all decision making and especially in times when we are afraid and have fear it’s not accessible by the prefontal cortex where we do all the cognition this I should do better that where we do thinking um so especially when we are in situation of risk and fear it’s not accessible and we are not as so we are really run mostly by our steam brain it’s very depressing then we have our own individual biography and then the situation when it comes to IT solutions I solution it is from the user perspective the user interface and the room a person lives in and then we have the psychogram so um based on genes your DNA and uh when it comes to trust building some people have the tendency to for them it’s trust building is much more easier for others is more difficult the mostly reason is uh based in early childhood experience and then we have have this unconscious conscious decision making process process so most of our decision making trust building very frustrating happens unconsciously so our brain first it prioritize then it does comp a complexity reduction Next Step compensation of information deficits especially when it comes to Ai and IT solutions we never have all informations which are being needed to make a good buying DEC decision or whatever and then very importantly the risk chances evaluation and um and Trust from a neuropsychological perspective means to skip this step and because the value of not analyzing the risk is higher right than to analyze it right because why it saves energy and this is the reason why we we love trust so much um we love to trust yeah because well from a neuros psychological perspective this is just only one discipline it saves a lot of energy and then we make a decision this whole process is tremendously influenced by 185 biases and horis STS which are horis STS and noises and so at the very end humans are not that good in decision making especially in this complex World surrounded by fear and complexities so um trust is the essence which drives relationships performance Innovation efficiency Effectiveness results successes as reduces direct costs and transaction costs it reduces M mistakes there’s so much more value to trust each other and to build environments where really could substantially build trust in each other and but what makes trust building so difficult it’s a global Geo geopolitical social uh social situation growing life and business complexities technological complexities um the speed of technological development the lack of transparency the lack of automated tools to monitor regular itary uh implementation the lack of efficient effective regulations uncertain uncertainties in data sources and data exchange and so for and so forth so it’s a very challenging time and especially when it comes to AI right incorrect results bio results manipulations physical harm psychological harm there are lot of reasons to be worried so how can we build trust this is quite complex uh very complex um the here I use the concept of the M’s pyramid of needs so at the lowest level following masslo here you have your basic needs eating sleeping living drinking I transferred it to building business models I love building new business models and there’s by the way there’s a lot of economic potential when it comes to security and Trust building um so business model needs a hard problem solving quality like information sa saves time Etc then it comes the next layer when it comes to AI correct non-biased training data transparency databases and data processes then we have the classical Norms like cyber gdpr and to AI we will probably will be trust or trust will be built by the process the AI is embedded in one example I came here by plane I don’t know all the safety regulations when it comes to airplanes and flying um even an airline engineer does not know all aspect how safety is being enabled and built in airplane but as a pass messenger I trust the process all the processes around how an a airplane is being built how does how it is being maintained how no how it is being operated and this also will be happening in the future when it comes to AI we have heard already a lot about standards I’m working in the standardization organization so we are looking at the content and um we also should really should focus on the governance systems around right to really ensure that the standards the regulations we are building currently are properly operationalized and um to enable something very important for us humans we are humans we like controllability we like reliability transparency accessibility predictability right uh because we do know how easy it is to manipulate us and especially when it comes to AI wonderful experiment um oh no it’s a different one um as soon as it was a diagnosis support system Healthcare and there’s a strong tendency to follow the advices of the results of the machine without questioning them there was one experiment in Social Services in Ireland and theyve manipulated an AI machine by purpose to give unfair very very unfair results and on the other on the other side of the keyword we’re sitting very very experienced social workers and in 87% of the times the very experienced social workers agreed to the very unfair decisions the machine made regarding the allowance of soci social care why yeah our brain does not like to work machine says externalization of responsibility so when it comes to trustworthy AI we really should spend a minute on how should we design the whole process and how shall we design the user interface so interacting with the user inviting him to use its brain to question the results um and also for us humans it’s we want want to play a role in this world and have a work satisfaction so I’m a huge fan of AI but a bit understanding of humans might be useful humans loves brains uh humans love brain yeah of course now um the brain loves labels also it’s a complexity reduction we don’t have to think it’s yeah it’s a uh information compensation process and um it’s very useful especially in areas where we don’t know have all information for instance how food has been produced was it organic or not organic and especially intensively used use labels are being used in medical technology and um yeah and then yeah especially when it comes to vital interest then we really love labels this has been intensively studied and analyzed especially in healthc care I do have to watch the time so and quality and it’s reliability communication has a significant influence on consumer and user behavior and uh label are wonderful example to communicate really the the the quality and gives really a freedom regarding pricing and it really raises the user ratios and uh they do have a value and that’s the reason why we are several organizations are busy building standards and labels we at the vde have built open standard this is the vde trust label by purpose we have used here the Nutri score um visualization because this is a very learned behavior and we analyze very clearly these five uh values accountability fairness privacy reliability and transparency and in different shades and there we just only look on the observable so there is no smart software behind we just only look at what is observable and for that we use vcio model so we have the first layer the values and then we have the criteria in this example documentation of data sets Etc and then indicators and then we go to the next level observables and all the findings are being assumed up to have to be to get a very specific rate why because the risk appetite very highly from the intended use of an i solution one example um a patient has for instance um bone cancer stage four which is highly highly serious and critical you’re probably probably privacy in such a highly serious situation might be not that important compared to reliability right and uh different use case for instance gaming they have privacy yeah or probably depending which kind of game right there’s privacy very very important and other values are not that much important because risk tide will really vary from use case to use case yep so and we have developed it with quite big companies like Bosch seens the tur SRP s um BSF kit and lots of Academia and we are already started working on um on version two where we also looking very closely at gen AI and really uh helping and this is by the way this is an open standard so you can download it for free on our website um also can use it for instance as a guideline to make better buying decisions for free no one will charge you you can use it as guideline for to build an AI solution um definitely and if you would like to have a certification we’re happy to get your call and um yeah it has a lot of ADV advantages it helps you easier implementation of the AI act um yeah it’s a it’s a quite concrete guideline for product development and we are happy for technical support no one is standing here so I have to finish so thank you very much and I’m open for your questions so really a great approach and great presentation about the trust and how trust is really important in all of the decisions which is covering uh the current implementation of AA act so it’s really really great pleasure to have your presentation thank you thank you very much pleasure being here and now we will go to really interesting case so how all of the applications could be empowered within the Europe so we know that a lot of innovation is really made in Us in China and Europe needs to step up so let’s give it a talk for orestes the director of Silicon Valley Hub of eat and he will try to cover so what Europe could do better thank you uh the topic is um it’s great so yeah I will really try so this is the right word so I’m representing the ATI community and uh the E Silicon Valley Hub so I have two hats and I’m working e digital a lot of complexity we’re working with you I think in the deploy eii uh platform and uh we do have activities on education and Innovation boosting startups and creative education programs so having said that I thought every time that we’re discussing on AI at least me uh I hope I found really useful before I jump to the to the main topic of my discussion to understand again thei to read the description I think the most of us here I think we have some lawyers but we have also Engineers I remember a professor was always telling me the half of the solution lies in the reading the problem so let’s look at what is the AI actually with other words so different definitions about what would be the future of AI before we see what is the definition I like all of them and they you see that they Contra contract dict each other so but is it is the end of the human era in order to understand this we need to understand what is AI and what could be is it new is it A New Concept no we’re discussing about AI many decades now what it changed and we will see this I will emphasize this later is the availability of data and the computation power that we now have and those two are the the only two things that uh consist in Ai No the algorithms the logic the mathematics behind are the things that are making the AI what it is today and the last two things we know them for a long time now so you can see how it how is the trend and it’s an important thing to know because we have already built the foundation of what is happening today and here is the question so what is this thing that we are all discussing and there are different approaches my personal I love the first one what is the AI I even read it artificial tell the science of how to get machines to do things that uh do in the movies and please hold this definition it will be relevant later this is actually how was it the the the definition in the AI act a very lengthy one you can read it I’m not going to read it for you and if you like the one that I usually use when we discussing about AI is the last one AI as it is today is an automated stupidity an automated idiocy this is how we need to understand it in order to understand the challenges that we put in the legal and the ecosystem in our lives super is it AI good yes it’s super good and here we have some very concrete and uh very valuable Sol Solutions already for those of you that are not very technical you the the right hand side you can see a little bit the mathematics and the model behind Ai and what it is and maybe connect what is machine learning and the other terminologies they’re all part of the family and on on the left side you can see some very good solutions that you already know but maybe I will list it image recognition style transfer synthetic image generation super resolution a text generation and where AI is good today but that opens the question is it good in everything no and I of course I forgot to mention sorry chess games AI is beating the humans already go very recently they also beat the AI beat the the world champion so AI is good and image recognition and I mentioned before and this is a very nice example they did a research analysis on this but with only 1% of the pixels available they were able to recreate the whole picture so there was not you see the true reducted images with only available 1% the AI was able to create the the right column exceptional AI is also doing well I mean I put this on purpose you will see it’s not actually true but it’s also doing well in automation this is a picture or this movie small movie of San Francisco uh the the the the company is called wio and has automated cars that are are up and running San Francisco if you go you can take it and you can go wherever but is it true that the auto it’s it’s very good ATO automato car no actually and this is why the wio is actually a pilot uh program in San Francisco there were also other two companies that were doing the same and because of incidents they sh down their operations so yeah it’s not as good as we as we think right now there are problems you see AI it’s some robots that we know they are failing to do very simple tasks H this is actually sorry it’s not a movement the turtle it was a very old experiment a couple of years ago now they’re a bit smarter but the AI couldn’t recognize the turtle as a turtle they thought that it was a a weapon this is what it is and you can see if another AI couldn’t understand the stop sign because it was a moving so I was understanding that it’s a different thing so AI is not as good as we things still and we need to hold this uh when we’re discussing about this and to not conclude too early that we understand it because as we can see in also in other applications AI is not doing well simple data analysis so super biased Pro uh programs that they are even though that you maybe you cannot understand it but they’re giving low risk to the white men and high risk to the to the blackmen only because it’s black why because the data behind this AI models was wrong that was the main the fundamental problem but this is still the case this is still a problem and there are also other Solutions I’m referring to this article you can go and look at it I have also limited time so I cannot elaborate a lot I’m giving you now some some ideas and AI it’s smart but not smarter than the human intelligent right now we also should not forget this so the right picture is super nice look what this smart guy did they took 10 mobile phones moved on the street and they blocked the Google AI so so Google Maps because they thought that the road is super busy so sheated the system the AI couldn’t couldn’t recognize that couldn’t realize what is happening of course and the the the right one is that the AI as I said the tomated solutions are not uh very well why because AI is not very good in solving very complex Pro problems so it’s good at chess because the chess has some specific rules but when it comes to interaction with many different uh things when we have multiactor problems AI is not good okay I establish something this is a very long I will I will need to to skip itth in order to save time for for the rest but this what what you should remember is that the AI as it is today is super data Centric and AI is not working like the human brain so we’re is and it’s only good as the data it learn from okay problems oh first of all my my slide is has a problem but in order to have thei that we see on the Matrix there are still things that need to be solved we like robustness the AI it is lack causality contrafactual predictions we we lack the ability to compete and to cooperate with each other even the AI system so we’re are far away from the real Ai and this is linked to what I was said we now what we have it’s an automated stupidity so but okay let let me summarize those things what is the AI today is data simple scenarios this is how AI solves problem s statistics algorithms gpus the computational power and the special knowledge sometimes on when especially when we apply it on the health sector Mobility requires this uh this special knowledge look at the second one we use Simple Notions that bombarded with a lot of data are solving complex problems and to me to understand what is the AI it’s the third bullet AI is creating algorithms that create algorithms that will solve the problem as I said AI is not working like the human brain we’re just train an algorithm to create the solution we’re not able to understand how it works actually and this is a problem for the future okay I spent the half of my time establishing what is AI let’s go now to the to the main ER topic of my of my discussion EU UI ecosystem look at the charts you see the different in the different countries but I will focus on the middle 6,000 European players on AI we have let’s say approximately 2,000 startups it’s again a big discussion what is an a startup and what is not I would call the ones that are have a a deep Tech solution you can compare it also with the other players you see the US China UK how strong it is and India look also what who is the actually the AI players it’s actually business not the research centers not the government so it’s the the the ones that are the actually players of the AI how how much we invest on AI not much we should invest more especially compared with the other continents and how much productive is efficient is our investment not much look at the second bullet only 10,000 million spent on AI in the Horizon 2020 resulted to 0.65 patents this is really low in the next in the Horizon Europe we’re doing better in the funding program and connected what was mentioned before and I really Applause the previous uh presentation the Europeans do not trust Ai and this is a problem not only that only onethird of the European companies are using AI this is a problem and this is why we have the AI act but I will come to this so again I don’t know why my my slide is like this really was better I promise you it’s complicated now of course to explain you what is happening here but in in a nutshell we have three times less funding for our startups and scal UPS we have three or four times Less scups in Europe our scale up density it’s very low compared to Silicon valy in US is this 185 9.5 in Europe have 3.6 in Israel they have 26.9 and the scale up investing ratio it’s also very low so we are not performing well ER in our investment in Europe I will I summarize this for you so we lack investment compared to both us and China we lack Talent because yes we are good in the we have very strong minds but they’re all fleeing to to us the most of them we’re doing well in regulation and this is important and uh we don’t have many unicorns which is actually super important problem for European Innovation be why because yes it’s good if we have startups it’s good if we have good scale ups but those companies are start returning the investment where the become unicorns to the ecosystem many of our good and promising startups in Europe they are going to us in order to scale up so we’re just handing over all the good ideas that we have in in us which is not actually a problem but seriously we need to to change the the balance we need to have good and strong scale ups and companies here in Europe so remember those things and uh thankfully a lot of the previous um uh speakers uh spend a lot of time discussing about AI Act and the governance uh scheme in Europe so you see a little bit what is the structure of uh the ACT I thought that would you find useful to see when we will have the first standards it’s it’s um late 2025 so there is a lot of work to be done before we really can Implement AI and also I see that they listed uh the governmental bodies or the governmental initiatives that will drive the AI two things are important to be said here two questions one is who is going to be the people who are going to be the people that will work in the AI office who are going to be also the director mentioned this but who is going to be the people what would be the mindset are they going to be the people that we want to have here in order to govern the AI this is is there still an open question and to me actually how much of the AI companies how many of them will be regulated not as many as we think only 30 40% of them could be heavily regulated here are the challenges and opportunities that we have as a European ecosystem I already mentioned a lot in the in the previous slide so I will focus on the opportunities the regulat framework it’s something that we need to embrace and I will come to this very shortly we need to understand that we have a lot of smmes as I showed you before and this is a good thing we need to work with them we need to connect them and see how we can leverage their flexibility and adaptability we have a lot of Rich culture which can be something that could really help us train better our AI I will give you very simple example you know that the Jud DDP was using the European commission document m ments in order to train the AI to speak in different languages so they use the commission papers which is I found very very very very interesting so this is something very well very good and we also can sprayhead the green and sustainable AI I see I see the time so I will be quick so what we what is missing how we can Empower AI in Europe we need to scale up the solutions I mentioned this before res skill and up skill we need to Res skill and up skill our people and our Workforce we’re doing well but we need to do more invest more and it’s already happening in the hardware and work more in the regulatory framework and why and this is last three slides I promise so I see the time we don’t need any more AI experts in my opinion we need to it’s good that we all understand the black box it’s good that we understand the social impact but we need to in order to really democratize it we need to develop AI in Europe and in order to do that we need to boost the creators the innovators this is a hot take and to adapt and operation agility and focus on people that are super talented when it comes to the governing of AI people that are really entrepreneurs people that understand the innovators this is the mindset we need to change a little bit bit our mindset and we need to understand this first the future of AI is thrust I agree on that an ecosystem of thust is a prerequisite why what I showed you before the fact that only 35% of uh of uh the Europeans trust Ai and only 13d of the businesses are using AI it’s highly problematic because our Solutions cannot scale up in Europe This is what it means we cannot have users so we need to change this we need to upload the commission’s uh initiative to to create the eii and all the legal framework because that will help us increase the trust rate in the AI and not only that to embrace it and create the ethical AI label based on the UI act that will be pivotal key for our future of AI so every European company when it’s a European it’s immediately unethical everyone knows this in the world in the world so they can really leverage on this and last last but not least we need to scale up I mentioned this already how we can do this we need to Har harmonize not only the regulation the tax policies to streamline the funding between the member states and the European funding there are different bodies that there are funding kind of the similar way we need to Strate this funding in order to increase the funding for the startups and to move from low and funding approach to the equity embracing the high risk High Gain mentality many many thanks uh that was my presentation I’m open to any discussion later many many thanks this thank you so much um we have um because um we have to be finished at in 15 minutes um we have to look a little bit on the time so one thing you will see now a video about a so-called AI Village uh it is not a virtual Village it’s a uh it is a former TV Hall where famous talk shows the Germans know Hans meiser and Elona Christen which were um done there it’s now a TV Hall um for AI it is situated near in Cologne and my good friend Alex Sigman will present it to you because it’s a very interesting take upon how to implement AI into smmes F from coal to AI this slogan was used by Minister President of the state of North Rin West faia hendrik vous in a recent announcement of Investments of computing Center projects in the region following this slogan I give a warm welcome to all the viewers and the whole European AI Forum from the AI Village our Innovative C campers for AI in the heart of Europe just outside of cologne my name is Alex Dickman I’m one of the project leads and I’m looking forward to presenting the AI Village to you since last year my team and I have been working on transforming this old TV studio that you can see here behind me into a collaborative Hub where companies startups and the public can experience develop and learn about a I the northw vestan government has committed to facing out liite mining until the year 2030 in a region where jobs and energy Supply depend on it this exit means drastic changes for the people their lives and culture funded by the federal government the AI Village is one of the anchor projects in the region to successfully accompany the change into a more sustainable a more digital future our goal is to establish the AI Village as a physical Hub in the region and to establish the Rin land as one of the leading AI regions in Europe we achieved this goal by building on three pillars experience AI develop Ai and learn AI our first pillar experience AI mainly addresses companies especially in the region through events presentations demonstrations we show them AI use cases handson and explain the added value for them next to this we are currently working on a demonstration Center in which the public can experience AI that you can touch once a company has understood the general value of AI our second pillar develop AI comes into play our team offers workshops free of charge to gain a deeper insight into the topic of AI and to identify specific use cases for the organization together with our large Network namely the German AI Association universities such as the r colog and research Partners such as FR Hofer institutes we then put together consortia to jointly Implement these a I use cases through our work AI startups do not only have the opportunity to gain valuable access to potential clients and projects we are currently developing an accelerator program specific for AI projects and we offer co-working and office spaces such as these as well as a growing network of experts experiencing Ai and implementing projects are a good First Step but they are of little use if employees do not know how to use these AI tools this is why we have our third pillar learn about AI our experts offer various AI courses to empower employees to identify I AI use cases in the organization to work with these AI tools and to promote them in their company with these goals in mind we brought AI closer to over 2,500 people at events in our studio here directly just in the last year in the same time we held indepth discussions on the topic with over 200 organizations and we are already initiated the first dozen projects we will continue this work in the coming years in order to establish the Rin land as a leading region for AI in Europe to achieve this I invite all of you to reach out to us visit us here in Cologne and realize projects together in the spirit of AI made in Europe thanks to all of you so thank you and now uh to one of the last points Before We Say Goodbye our friend P from digital Poland well do we have slides for our friend yeah this is my uh it is my uh privilege uh on behalf of the whole European a family to announce that for the first time we are starting a project which is called European AI Awards uh you know you just learned that there are so many ideas so many scientific breakthrough so many startups investors coming from all over the Europe and the problem is that we as a Europe when you compare to us we don’t have just just one Hub like Silicon Valley and maybe New York we are really distributed which means there’s a lot of in Innovation coming from Paris from Berlin but also from Lithuania also in Poland and probably you’re aware and we are aware that there are great new companies like Mistral from France Alf Alpha from Germany and a couple others like 11 Labs from Poland but we uh discussed uh with within our family that we need to present the whole ecosystem because one of the main challenge for example when you are startup you’re looking for investor or you are looking for a scientific breakthrough and you you are not aware that there are good investors in in in our ecosystem that’s why we decided to announce the European a Awards we’re going to uh open the application form at the beginning of uh July uh we will announce the next uh the first winners in uh December on our 10 edition of the European AI forum and the the European AI Awards is splitted to five categories first of all we will uh award the best AI investors what does it mean we are looking for business angels for private equity for for Venture Capital who invest in best AI startups but also in R&D then we’re going to also award AI businesses AI businesses are the best use cases the best scenarios where AI was applied so we will be looking for a successful Corporation who will share the uh the for example the use case and the successful implementation of AI for example together with startup then the third category is about uh spin-offs and um applied AI so as you know there are many interesting scientific breakthrough on universities but they don’t get investment so we’ll be looking for best spin-off that came from universities the fourth categories is the ecosystem enabler which means we’ll be looking for such projects as as the uh AI Village but many others which brings together the whole community and Build Together the the AI within Europe and the last but not least of course is the AI policy because as you know uh we live in Europe which is more regulated than the us we’ll be looking for best policy maker best government who do the best AI strategy best AI implementation to unite the Europe and streamline the the implementation of AI in terms of countries we are looking for countries for for eligible countries so we are looking for organization initiatives startups entrepreneurs who lives in Europe but also who establish their initiative in Europe uh as a Europe we doesn’t only mean European Union some 27 countries we include Switzerland we had a guest from from Switzerland Balkans uh also UK the only countries which we exclude from let’s call our Europe family are the obvious evil ones which is Russia and Belarus uh but we include of course our friend from Ukraine so the whole Europe is invited to send the nominees at the beginning of the uh 1 of July and I’m really you know uh really happy that it’s it’s our first edition who will present the whole ecosystem thank you thank you yeah we are really looking forward to that so uh people out on the internet on YouTube but especially also here thank you for um having joined quite a long session of the European AI Forum uh I know it’s a lot uh but you have also of course the possibility of re-watching the most interesting Parts on our YouTube channel uh please like And subscribe on LinkedIn yeah we’re really looking forward to that one and uh let’s stay in contact I want to thank Lenas and the um artificial intelligence Association of Lithuania for this great organization woo Lenas thank you we wish you all the best from CH Lenas and priot see you in six months in another location in Europe have a good day and goodbye thank you