 Efo'r cymhiliad am ydydd y gweithio, ydy'n hynny, yna gweithio, a'i gyd ymydd y bydd yn brydol. Yn nghymru, mae'n gweithio i'r ddefnyddio, mae'n gweithio'n 3 oedd, 2 oedd o'i gweithio'n gweithio, sy'n gweithio'n gweithio. Mae'r rhai oedd yma yw Maddy. Mae'n gweithio'n gweithio'n gweithio yma yw Maddy, ac mae'n gweithio i'ch ddefnyddio'r gweithio. Felly, roeddwch i'n gweithio Maddy. Mae'r ddaeth i'w gweld i'w gweithio'r ddifenol gyda'r ffons iawn, a ddim yn ychydig, dwi'n gweithio'n gweithio'n gweithio, a ddwy'n mynd i'n gweithio'r ddifennol gyda'r ddifennol. Dwi'n cael ei ddweud a chym ni, Maddie. Mae'n ddifennol yn ysgwrddio. Maddie Delvo yw Lachanbrg. Mae'n amser yn 2014, ac yn gweithio'r yrhyw ymlaen o gyllideb. bod gyda'r Minister for Education i niw feddwl yn edrychiol a'r Minister o Cymru Cymruildedaryllol, ymweithio cyfaintol. Wnaiedd datblygu'r rhaglen o'r rhwng ddweud â'r dros, yn gyfweithiaid am unrhyw deallu bryd er mwynhau ar gyfer y baith a'r rhwng oedd yr argynoddau cyfaintol i yn wedi bod yn ymwy� ymdwyllgor ar gyfer yma am y cyfan opeth, fydd o'r cyffredigau copying, robots, data and the age of artificial intelligence. This report has been very well Mae'n ddweud yw'n gwybod yn ysgol, mae'n ddwy'r ddweud yn 2017, felly mae'n ddweud yn y gwybod am yr ei hyn yn ysgol, mae gennym yn gwybod hyn yn ei ddweud am y dyfodol, yn y parlymyd, yn y parlymyd, ond yn y cyfnodol, ac mae'n ddweud yn yr ysgol. Yn y cyfgrifennidau yw, wedyn ei wneud o'r gwybod o'r rwynt i'r cyfwyr. Mae'r unrhyw yw'r ysgol yn eu ddweud i'r gwrs, I know Maddie is going to address this, are the creation of a European Agency for Robotics and AI, a new legal framework to accommodate the wider integration into society and a consideration of a universal basic income. So Maddie we're delighted to welcome you here today to discuss these issues but also to look at the opportunities, the threats that are brought on by a rapid advance of AI and robotics and really to ask the question, what is the prospect for Europe and how can we become in Europe a global leader in this area? So Maddie we look forward to your presentation, thank you very much. So good afternoon and I am very pleased to be here. I like Dublin also it was quite difficult for me to manage to come because I was supposed to come in February and then Dublin was blocked, snow and yesterday there was a storm over Boston so my plane had three hours delay but I want to thank you for the invitation. So first I would like to say that I am not a scientist so I am what I would call an informed citizen who is interested in the topic and we started to work on robotics in 2015, it took us two years to come to a report which is a lot of time but you know that the European Parliament is not the right of initiative so we cannot propose a regulation or directive but we can make recommendations to the commission and ask the commission to come forward with initiatives and with proposals. And that is why it was a so-called initiative report which makes our part relatively easy because we can launch ideas and we don't have to implement them but it was a very interesting work and I would say if I had to do it today maybe the report would be more granular, not so general because at this period artificial intelligence was not yet it started to be a big issue but it was not yet in the hype so that's why the title is of robotics but of course we touched on artificial intelligence. So first I would like to say that we tried to have a balanced report so to show that artificial intelligence and robots are useful to humans that they can serve humanity, society and we want to be positive to innovation and also to encourage European industry. But on the other hand we must not close the eyes as their challenges we have to look at and to discuss because personally I believe that artificial intelligence and robotics will be successful if there is trust of the citizens and the consumers. And so we have to try together to find the right solutions in order to realize the potential of artificial intelligence and I want immediately to say that I am not a fan of science fiction. I know science fiction is very popular and these ideas that there will be super humans killing the humans for me this is not what is really happening. And I'm afraid that I am very frequently in conferences where you have these discussions on the next generation where humans will be combined with robots and artificial intelligence and they will be super humans. But I think this discussion tries to hide the real challenges that we have to face now and this is what I think we have to concentrate on the existing challenges. So I want to take to tell five challenges and the first one is about safety and security because if there is not a maximum level of safety and security people will not trust. There is a big fear of cyber hacking, cyber security and we know that 100% of security and safety does not exist that does not exist but I believe that this aspect could be enhanced. I once listened to one of an expert who said that certainly we industry and politics were not attentive, not sufficiently attentive to this aspect of security when building the infrastructure and that now we have to pay for this. And I am very glad that the commission decided that cyber security would be one of the priority in research because for the moment there is nobody knows how really to organize the cyber security. So we call of course for safety per design for security and for very high standards which have to be defined at international level but we need these standards also to, this is of course not hard law but if there are high standards we can impose these standards for every entry to the European market. Because for the moment safety is controlled so you can be certain that product with a label safety will not explode or normally not but there is not such a label for security and so this is the example of the TV that listens to you which should not happen. If we had a control of security so and one important element where we insist on is on testing because these things have to be tested for a very long period to be safe and it is important to define the standards early because the industry has to adapt. They have to know what standards have to be implemented and if you talk about internet of things of course it's the weakest element in the shame that will be hacked maybe your toaster or your coffee machine and so if we have no standardization they will be on the market and we will use them. Okay the second challenge is data protection and privacy. Even if we have now the GDPR which I think is a masterpiece of the European co-legislator because it's a better protection of privacy and well we will see how it will be implemented because 28 member states have to implement and regulators have to control which is also a big challenge. Nevertheless for me this remains the most difficult issue because artificial intelligence and robotics do need data and they will collect an enormous amount of data and especially robots I think of care robots or medical robots they will collect a new kinds of data. If you have a robot in your house this will change your life and this will change the quality of the data that will be collected so we have to find answers on how to protect first privacy but also we know that we need data so they have to be available but who has access to this data which is the biggest challenge for me. Does it belong to private companies who do not give access to others if data is gold or the petrol of the 21st century then it's of course important to know who can exceed to this data and then there is a big issue of the bias in the data. Of course I didn't meet a single operator or industrial who didn't tell me that they do everything to not to have biases in the data and that there is a control and but I haven't received a satisfactory answer on how they deal with this issue because well we know that. Industry is very male dominated and so there are not many women and we know that the data are very frequently come from white man of a certain age and we see this in face recognition but also in voice recognition because voices are easier. Our devices recognize more easily the voice of man than of women and of children. It's very complicated. Other people is also very complicated to recognize so there you see that there is a bias in the data and this has of course an influence on if many decisions are made by machines then of course it is a big issue. I know that research is going on on this issue but we are not at the end of the story. At then something I learned last week so we normally in Europe we think that we lost the battle of the data because the data of personal data at the big US companies I don't speak about the Chinese companies who have enormous amount of data but in Europe we have an enormous amount of public data which are not available. Normally not available for companies but in business to business Europeans are leaders in artificial intelligence so I think we could focus on this part of the business not business to consumers but business to business where you can use also public data. There was also a call of the French government to open public data to research but this if we want a European market, single market then of course it is necessary that data are compatible and every member state has or in a member state every agency has its own design so this will need a lot of investment and political will to be successful. For the third issue is about liability which was the core of our report because the question is who is liable for the action of a robot. We agreed that it should always be a human there must be a human in the loop but it will be complicated to allow responsibility between the designer, the producer, the network operator because there is a network, the data provider, the owner or the user so this is a very complicated issue. We in the report we took the view of the consumer if damage occurs to a third person there must be a reparation must be made and so in the recital we called on the commission to repeal or to amend the existing product liability directive which normally should apply in this situation. So the general principle is that the producer is liable for an accident caused by the defect of a product so the victim has only to prove the damage and the causal relationship between the product and the damage. The product is considered defective when it does not provide a safety which a person is entitled to expect and this will be the difficulty with robots and mainly specially self-learning robots because if they take the data from the environment and if there is an upgrading, a regular upgrading which happens with all our smartphones we normally receive a call to upgrade. And if you don't do it then is the user liable because he was negligent or was it a default by the producer so these questions are not clear and so you have a multiplication of intermediaries that make it also difficult to identify as a liable producer. What is the reasonable use and so on. So we suggested that the commission should make a proposal and I know that the commission is working on it. They have been working on it for four years now without coming to a conclusion because there are two schools. The people saying that the product liability is perfect but now we saw that in the last communication of the commission, the commission agreed that they would give a new interpretation of the product liability directive. So this is a signal that it is not so perfect and that there has to be a change made and mainly because the definition of a product was this directive dates from the 80s and of course there was no discussion about algorithms. So the question whether an algorithm is a product or not a product is not clear and so we are expecting this interpretation in the next months but there will be not a new proposal. We could also have a different, there is a school of thoughts proposing a different approach, the risk management approach where the focus is not on the person who acted negligently but on the person or the company best place to minimize the risk. That means that you accept that there is a risk but as soon as you discover the risk you have to repair it. So this is of course also an approach. Personally I would prefer this one because with an evolving technology it's very difficult to define the standard that is expected because normally if I take the example of cars you use your car for five, ten years. And so if the standard was correct when you bought it maybe five years later it is no longer, it doesn't fit anymore. So this is of course a big issue. And then we proposed an insurance scheme at least for some robots and this of course implicates that we have a classification of robots because you need to define different classes of devices and for this first we need the insurance industry. And when we started to work they were not very interested in the topic and of course I understand because they have, there is no statistical data available and insurance works with data so they don't know at what level to insure and we have for the moment at least no classification of robots. There are different definitions but there is no classification of robots so this is a big work that has to be done and we were very much inspired by the existing insurance system for cars because there you have a compulsory insurance and with a compensation fund behind so I think this could be a model for some type of robots. And then we were very inspired in our discussions because the working group worked for one year and a half so we listened to a lot of experts and the next generation of robots will be self-learning and so the link between the producer and the robot is less evident and then it will be even more difficult to establish liability. And that's where the idea came that maybe we could think about a new personality, the e-personality and this raise of course an enormous discussion, very passionate, not only in the parliament but in the large public because there were some very, very right wing people who pretended that we wanted to give human rights that we would want to consider robots like humans. This rights and duties and this was a very heavy campaign but nevertheless the idea that at least we should discuss this issue remained in the report. It was adopted by a majority of members of parliament but the commission did not reply to this suggestion so I think the debate will go on but not on the level of the institutional, not on an institutional level. And of course I understood also that this could be an element to take responsibility away from the producer because if he is no longer liable for something then he is less careful I imagine to put something on the map. This is a strong argument against the e-personality. Okay, then I come to our fourth challenge which was which is ethical principles. So we have of course values and principles that are enshrined in the Charter of Fundamental Rights and UN Convention for Disabled People. I think we must not neglect this because I believe that robots will be very helpful for disabled people. If you look at proteases and so they are fantastic achievements but this will have a consequence on what we consider as healthy and not healthy. This will raise a lot of philosophical debates on eternity, how young, how performant you have to be and are you responsible for getting sick if you have a lot of means not to get sick so this will be a very complicated issue. But non-discrimination, dignity, autonomy, freedom and this principle must be respected by robots who interact with humans. So we propose that we should have a code of conduct for designers, people, engineers but also for industry and for the users. How do I use robots and the creation of ethical committees in institutions because even if you define principles you have to go down to a practical situation. I don't speak about the car whether you should kill the pregnant woman or the old man but if you are in a home for elderly people and so the robot is supposed to serve the person but if the person has dementia who then decides what to do? If he or she refuses to take medicine, must the robot force it? So all these questions have to be dealt with and they are very, very practical. And so I am very glad that now these ethical discussions are really in the public and so you know that in the communications the commission proposed that still this year they will come forward with the expert should elaborate a code of an ethical code. But there are so many bodies working on ethical principles now that we have to be careful that we will not have a zoo of different codes so we need some harmonisation or collaboration between the different bodies. And my last challenge would be the societal challenges and mainly jobs and education. So we know that well I have one conviction that artificial intelligence will change all aspects of our life and it will have implication on all the professions, all of them. And so some, of course some jobs will disappear which is I think normal with technological development in the past there were professions that no longer exist today and new professions arise. So there will be losers and winners and I am not very worried about the winners but I think nevertheless that we should take care of the losers because one thing is to say that maybe there will be in the whole no loss of jobs but if I am a person who loses my job and I have no perspective to find a new man, then this is nevertheless a big worry and for me as a policymaker I think the challenge is how can we organise a society in harmony because social peace is I believe our biggest good. So the answer is education of course but I was in my former life Minister of Education in Luxembourg for nearly 10 years and I know also from discussions with my colleagues that education system are very slow to change. And so this is really the big challenge and I believe that we should in the meantime I have the conviction that we should pay more attention to lifelong learning and organise in a better way and give incentives to people to profit of the opportunities of long life learning. And because statistics there also show that the better your higher education the more you use lifelong learning opportunities and it should be the reverse that people who had not a good started education should have more opportunities to upgrade their skills. So this is of course I don't know how to do it but I believe there should be incentives and that's why we came to the proposal that we should maybe give an income to people who have no job but give them the opportunity also to have their mind free to use these opportunities of lifelong learning. But this I have to say it was in the draft report it was not adopted by the parliament. The parliament refused this idea. So there are also discussions on social protection because our proposal was that we should be prepared for different scenarios. If everything goes well then of course we don't need to change many things. But if there is a shortage of jobs then of course our financial systems, our social security systems will be in difficulties and so we should prepare for different issues and certainly have an improved monitoring of what is exactly happened on the job market. But I know that studies are now ongoing and of course it's too short to know exactly what will happen and by definition the future is unforeseen. So but at least between MEPs we could agree that we need more expertise and that's why we asked for the creation of a European agency where the expertise could be brought together with scientists, computer scientists but also lawyers and experts on ethical questions. Because I think the debate on what is the right principles to be enshrined in robots should not be left to engineers but that it should be a broad discussion. So the commission told us that they don't want to create a European agency because there are too many agencies already. And while we also can live with this competence given to an existing agency and I don't give up hope that we will have it one day. And then the other thing is that we need more debates and better information of the public because when I go to a broader audience people speak about Terminator but not about Toaster. And I have to confess when I started to work on robots I had also this imagination and when I saw what a robot is for the moment able to do that's not so fantastic. So I think we must tell people what is possible and what is not possible in this domain and what they can expect and what they must not fear. And so we need an informed debate and an inclusive debate not only between experts. So I, and this is why I am very happy when debates are organized around Europe and I have to say that now we are for 2018 there are many debates taking place and I see a big participation also of experts and policy makers in this. So you know that in April the commission had a communication finally where they propose a strategy for the union on artificial intelligence. I think this is a step in the good direction. The commission first promised more money and I hope that it is not only a promise but that it will really happen because we need investment when you compare what the amount of money that are invested in China, in the US, in Japan and in Korea in the research and also in the industry of artificial intelligence. We have not enough money so this will be made on the future budget of the European Union but also on the involvement of private sector because the big research is made by private enterprises for the moment. When I see the money Google, Apple invest in their research and when I look what is happening in public labs so this is quite frightening. And that's why I think we need these codes ethical charters to be adopted as soon as possible to have a level to make clear what we don't want and what we want. And then we have this EU alliance on artificial intelligence and I would invite everyone once the platform is launched to participate in these discussions and we are waiting for the interpretation of the product liability directive. But I think the most important would be the involvement of member states that finally member states do not have a national strategy for artificial intelligence but that 28 or at least 27 will have a common strategy for artificial intelligence because I believe that we can win the battle of artificial intelligence if we stick together and if not everyone. And this will be if it is done the right way this will be beneficial for humans and then we have to pay attention how we will organize the distribution of wealth because the accessibility who will have access to these new devices and how will we finance this, this will be the big issue after. But first I think the first thing to do is to remain united and or to keep together and to show that Europe remains competitive in this issue and but the societal debate should not be neglected so I was too long and I thank you for your attention and I would be delighted to have your questions, remarks, critics, everything. Thank you.