The Europe Strategy Research Institute at University of Public Service, the Digital Success Programme and Ernő Schaller-Baross Member of the European Parliament organized an online conference on the 16th of March about European AI policy. The event had several guests from V4 countries, legal, technological and ethical experts on AI.
The conversation was held during the European week of AI 2022. Not to mention the fact, that the negotiations on the AI act, proposed last April are ongoing and the European Parliament Special Committee on Artificial Intelligence has drawn up a report on the social, technological and economic impact of artificial intelligence. The event had a dual purpose, on one hand it challenges the evolution of artificial intelligence and its regulatory aspirations by posing multiple questions from the AI ecosystem of Central and Eastern European countries. On the other hand, the event, as a bottom-up initiative, aims to strengthen the CE regional cooperation between different sectors to develop the future of AI.
Ernő Schaller-Baross outlined the importance of strategic thinking on AI, that drives to Europe’s digital sovereignty, and develops the global competitiveness of the integration. He added, that, the new legislation is a combination of the first ever legal framework and a coordinated plan. He added that the EU must act jointly, to ensure the ethical and safe use of the technology, highlighting the value of the proposed AI Act both for Hungary and the region. András Levente Gál, Professional Leader of Digital Success Programme, and CEO at National Data Economy Knowledge Centre explained, that the modern public administration has three pillars: clear data, useful AI, transparent and intelligible network. Based upon the pillars we are able to implement a paradigm change in the public administration. Finally, Bernát Török, Director of Eötvös József Research Centre said, that for the legislation of AI, a close cooperation is needed involving many players of the society in order to find the right solutions. He also pinted out that the regulations need to reach the social impacts and public values as well, because the AI appears in all different spheres of life.
Following the welcoming remarks, Irina Orssich (Head of Sector AI Policy at Directorate General for communication networks, content and technology of the European Commission) hold the event’s keynote speech. She answered four questions, about the legislation: why, what and how is the EU regulating AI and how does it foster innovation. According to the EC’s priorities, avoiding overregulation, they do not want to regulate all AI, just mostly the high risk cases and horizontally, not sectorially. The AI Act proposal aims not to regulate the individual use, but the AI itself, prior to entering to the European market. In this case, there will be criteria fort the AI system with regard to quality management, systems risk, human oversight, then the information to be provided to the user and for cyber security. Finally, it’s very important that all the legal framework goes hand in hand with a number of measures to support innovation. So, the EU does not only want to regulate, but in the same time it wants to promote its values.
The next speaker was László Boa, Professional Leader of the AI Coalition Hungary. Mr Boa started from the basics of AI in IT perspective. Basically –he added- there is some kind of input, a decision and an output. Firstly, the input is the data and it’s essential (for example for the training), but there are lots of security and technical questions around this topic. Secondly, one of the most important aspects of how the decisions are being made by AI. There are two options: the system can store the data in every decision (for example, in the case of online banking) or not store anything (chatbot). Finally, the output can be a single yes or no, or a complex identification. The presenter listed some examples for different types of AI, such as invoice verification (character recognition, classification), medical image recognition (detecting diseases and symptoms based on computer generated medical images), infrastructure protection (surveillance cameras, object detections), autonomous drones in agricultures (disease recognition, spraying), autonomous vehicle (radar and image recognition), natural language understanding (text generation, Q&A) systems.
The next speaker was Anna Zsófia Horváth, legal expert from CMS CEE presenting the ethical and security dilemmas of AI. Horváth highlighted, that technology has to be built based on human-centered approach, where the trust is the essential factor, that can be established in each stage of the AI. The system needs to be lawful, ethical, secure and safe. The training data should be large and representative enough, otherwise serious implications may occur. One of the essential questions of who has the final say or how much decision making power relieved to the machine. Because of these issues, a human oversight or a human intervention part is also an important question. Besides, the human judgement is worth a few words, because the AI system’s results would somehow directly or indirectly influence the human decision making. In addition to the horizontal framework, we should also identify situation based or application based issues- added Horváth. For this situation based approach, there are several examples: in employment or economy (in HR or in advanced industry), autonomous systems (self-driving vehicles, drones, weapon systems) and deep fake technology, or the use of AI in judiciary system. We can see, that the AI appears in various sectors, therefore it’s important to think about decision making, especially in the military sector. The frameworks have been developed to tackle these ethical issues. There are main principles to support the human centered approach with the respect of human dignity and human autonomy, also the transparency and explainability. According to the legal expert, other inevitable topics are cyber security and data protection. If, the goal is to assure a cyber secure rollout of an AI system, the way has to be paved by providing assistance and guidance on how to concretely move forward with cyber security.
The next speaker was Jana Novohradská, who is the representative of the Slovak Republic at UNESCO. Ms Novohradská presented about building out European trustworthy and excellent AI ecosystem. AI represents an automation, outsourcing and the decision-making support said the presenter. AI system being deployed for example in the private sector, and in the financial sector too. She highlighted, that there are a lot of regulatory, technical and ethical challenges. The answer, to all these challenges, is “to start actually doing things in practice, by cooperating with the private sector, public sector and civil society”. She outlined, that the digital humanism is a new positive international movement. “Our digital future is up to all of us: how we set the boundaries, direct the change and this is where the trustworthy AI ecosystem becomes very important”. There are many good practices and recommendations in terms of AI ethics, such as European AI ethical board and AI ethical officer. She was focusing on the AI egalitarian policies to support more women in AI. The UNESCO data shows that currently only 22% of AI professional are women. She closed her presentation with the need for digital public sector to come up with an innovative toolkit to support all these new challenges and functions.
During the roundtable discussion, the panellists discussed their views on women in AI. The first speaker was Dóra Petrányi from CMS. Dr Petrányi highlighted the importance of certain European values. Through the example of GDPR she demonstrated that the same doubts emerge at AI regulation, such as the fear of competitive disadvantage for European companies. Regulation can be very important because it can actually show a regulatory certainty, and can create a safe environment. On the other hand, regulation can serve as a boundary for innovation if it’s too strict. “That is the balance that we always have to look for.” She agreed with the values of the European AI Act proposal and the European efforts to regulate AI. She added that it will be extremely difficult to regulate such a big creature as AI on a smaller regional level, not globally. Taht is why the UNESCO’s principles are extremely important and probably to be followed by more global organizations, said the presenter.
The next speaker was Jana Novohradská, who highlighted, that in terms of women in AI leadership, there might be an active resistance. Sara Pollak, founder of the Painoeers Lab spoke about the low level of digital literacy in artificial intelligence. “There is a huge gap between the understanding of AI and what it can do for us.” According to the speaker, to develop AI, we should activate the resources that we already have and making sure that the right information, product or the right system gets to the right recipient. She thinks that when we talk about the AI, the average person feels disassociated with the technology, especially, because of the lack of education. “Artificial Intelligence can really help us with the transparency, it can give us a kind of objectivity that we need. Explainability in terms of ethics is a big point to focus on in the coming decade.
Answering to the questions received from the audience, Dóra Petrányi added, that culture is a part of who we are, therefore cultural differences influence AI, that’s why we need to tackle them. Jana Novohradská outlined that the cultural dimension is the human dimension. Sara Pollak told, that giving people privacy, regulating data, allowing innovation, are all becoming very difficult. There are people, who start to use technology in an evolutionary way. She also added that everyone needs to be as educated as possible in order to be able to survive in the digital jungle as effectively as possible.
Ferenc Kása, professional Diplomat for Data and AI, at National Data Economy Knowledge Centre concluded the event with the following thoughts: “today we are trying to prepare a proper regulatory framework, on the national and the European level as well, and we can see the global effects in UNESCO principles and guidelines. “We are still trying to guide ourselves through this math course by the new technology, called artificial intelligence.”
Daniel Fekete and Tamara Tuza, Mentor Program, Europe Strategy Research Institute