Regulation of Trustworthy
Artificial intelligence (AI) brings numerous benefits to individuals, businesses, and countries, e.g. when social networks recommend friends, when algorithms perform transactions on stock exchanges, when AI helps law enforcement agencies predict crime or the military search for potential terrorists, or when AI assists the elderly with daily routines. With the Europe-wide initiative on co-operation in the field of artificial intelligence, which Slovenia joined, the European Union has outlined its efforts to use big data, machine learning, and artificial intelligence in order to tackle a number of inherently social problems: poverty, climate change, social inequality, and criminality. These developments in AI must take place in an appropriate legal and ethical framework, which will reinforce the existing levels of respect for the human rights and fundamental freedoms of every individual and vulnerable social groups and consolidate legitimate democratic processes.
The targeted research project (CRP) Human Rights and Regulation of Trustworthy Artificial Intelligence (V5-1930) is funded by the Slovenian Research Agency, the Ministry of Foreign Affairs of the Republic of Slovenia, the Ministry of Public Administration of the Republic of Slovenia, the Ministry of Education, Science and Sport of the Republic of Slovenia and the Ministry of Justice of the Republic of Slovenia.
Our research work is divided into three parts. In the first part, the Atlas of AI Usage, we analyse the use of AI in social domains and social subsystems where AI can affect the foundations of democratic social order and the rule of law. In this part, we also examine, among other things, existing uses of AI in public policy. Based on case studies of good practices of AI use in public policies abroad and analysis of knowledge from existing platforms, we aim to formulate guidelines on how to successfully transfer and implement good practices and knowledge in the sphere of public policies in the Republic of Slovenia.
In the second part of the project, AI Effects on Human Rights and the Individual, we aim to develop qualitative analyses of the use of AI in different domains. The two key impacts of AI addressed by the project are the impacts of AI on fundamental human rights (legal aspects) and thee effects on subjectivity (psycho-social aspects). The starting point of the project is that the implementation of AI should be based on respect for fundamental human rights and fundamental freedoms, democracy, the rule of law, and ensuring economic and political stability. The diverse examples of existing use of AI show that while there are many benefits, new systems enhanced by AI can also lead to violations of fundamental human rights. Threats are domain specific – e.g. insurance, banking, justice, policing – but some are common to the operation of AI. Examples of the use of AI show that it poses a number of legal and ethical challenges. At the micro level, AI applied in decision-making systems about individual rights and obligations affects many individual human rights, such as privacy and data protection, the principle of equality and social justice, individual autonomy and dignity. At the meso- and macro levels, there are impacts on the functioning of democracy and an increase in economic inequalities, leading to social classisation of the population.
We are aware that AI enters a specific socio-cultural-political space, but it is also a product of a specific socio-economic situation, which it also helps re-shape. Therefore, AI can be used to enhance “negative” social changes, for example, by increasing economic inequality at the global level, expanding socially disruptive and hostile policies, or as a means to prevent such social change. In the framework of the project, we aim to analyse legal dilemmas in several legal fields and identify cross domain issues related to the use of AI in decision-making processes, such as the following: the non-transparency of decision-making AI systems; quality and biases in data; selffulfilling prophecies stemming from over-dependence on past learning cases; the danger of data maximisation; power imbalances resulting from the use of AI; the threat of data determinism and discrimination; chilling effects; and echo chambers resulting from personalised services.
In the third part, Regulating AI and AI Governance, we place the previously analysed uses of AI in an ethical and legal framework. In this part, we examine the different ethics and values we have in mind when we talk about ethics in AI. One of the central themes we will deal with is the protection of personal data, as AI has experienced a renaissance at a time of change in the intensity and scope of the collection, processing and access to personal data. The fundamental principles of personal data protection in the EU were developed in the pre-digital revolution era and are often at odds with the realities of data collection and processing in the era of big data and AI.
As AI is a global change that cannot be regulated at national level alone, issues of AI accountability, transparency and human rights implications are already being addressed in international fora, e.g. OECD, UNESCO, G20, EU, but a comprehensive AI governance model has not yet been developed. International cooperation can ensure effective global governance, but at the same time it is important to recognise that in the field of AI, rapid technological developments and information asymmetries between policy makers on the one hand and the community that develops and uses AI on the other seriously undermine the ability of all stakeholders involved to effectively address the issues that arise. One of the issues we are addressing in this project is how to build a comprehensive and robust global governance regime for AI, following the example of good governance models in other fields. As effective and accountable governance can only be implemented through well-developed indicators of AI development, one of the objectives of our project is to establish a roadmap for a single information resource that would offer a comprehensive view of development trends within the complex field of AI.
RELEVANCE TO THE DEVELOPMENT OF SCIENCE
The research project will contribute to deepening scientific knowledge in the field of AI in Slovenia and thus help Slovenia position itself as one of the countries at the forefront of the issue on a global level. Due to the interdisciplinary composition of the researchers participating in the project, several scientific fields will be enriched, most notably criminology, law, philosophy, psychology, public administration, global governance, international relations, as well as computer science and informatics.
Aside from contributing to the evolution of several scientific disciplines in Slovenia, the results will be relevant for the European as well as global academic community. The questions that we aim to address are clearly of global importance. The articles published, the conferences to be organised, and other deliverables produced will contribute to the exchange and dissemination of knowledge on a wider basis. By disseminating published papers as well as by attending and organising international conferences, our goal is likewise to enhance international cooperation with Slovenian and international experts. Networking and the maintenance of international scientific networks can then be utilised by cooperating in consortiums when applying for COST, Horizon 2020, and other grants. We also aim to transfer the acquired knowledge to our students in Slovenia and abroad where we lecture, i.e. faculties of law, the humanities and social sciences, social work, public administration, and security sciences.
By mapping AI usage and development in Slovenia, WP2 (“AI usage in various domains”) will substantially contribute to legal knowledge in criminology, as well as in various fields of law as our goal is to point to clashes that may occur in several separate legal disciplines as well as to question the suitability of the following legal regimes: competition law, privacy law, personal data protection law, consumer law, criminal and police law, and electoral law. In WP3 (“Analysis of AI usage in public policies”), we aim to contribute to the body of public administration knowledge by identifying AI solutions that will strengthen the legal and social state and enhance the quality of life of Slovenian citizens. Furthermore, sustainable AIbased solutions as well as smart solutions that can enable the inclusion of vulnerable groups of the population will be sought out, along with case studies of good practices in the use of AI in public policies abroad, and in existing platforms and start-up companies. Within WP4 (“The influence of AI on human rights”), advances in scientific legal knowledge will be made by analysing collisions between existing legal concepts and doctrines and human rights. This will likewise by achieved by attempting to answer questions regarding the common issues that may exist as regards using AI in decision-making processes, such as the non-transparency of basic decision-making parameters, the information bias of societal data, which is always socially and culturally biased, black box effects, self-fulfilling prophecies, the occurrence of a vicious cycle in AI predictions, issues with outliers, the danger of data maximisation, problems with poor datasets, the imbalance of power that occurs due to AI, the dangers of data determinism and discrimination, chilling effects, and echo chambers for personalised services. WP5 (“Psychological issues of AI”) will aim to answer the question of how AI can be of use to improve individuals’ psychological well-being as well as what the plans are to use it in mental health professions, which will contribute to the body of knowledge in the field of psychology. The findings on how the use of AI affects people in their daily lives is furthermore not relevant only for psychology, but a valuable piece of information determining the direction AI should take in various scientific disciplines. WP6 (“Ethical implications of AI”) will address the issue of having various ethical theories that could potentially serve as a foundation for AI, which may offer very different interpretations and attempt to answer the question of what kind of ethics we have in mind when we talk about the ethics of AI. The research conducted within this WP will contribute to the field of philosophy and law, with the former serving as a basis for research in numerous other scientific fields. Moreover, analysis of the ethical implications of ICT in computer science has become an extremely active research area as attempts are being made to design ethically aligned AI systems. WP7 (“AI and Privacy and Personal data”) will contribute to the field of law, more precisely, privacy law and personal data protection law by analysing the compliance of the fundamental principles of personal data protection with AI and assessing the legal position of the individual in automated decision-making. By creating recommendations for policymakers addressing questions in relation to building a coherent and reliable global governance system of AI, which is the goal of WP8 (“Global governance and AI: the positioning of Slovenia”), scientific knowledge on global governance and strategic foresight will be advanced. At the same time, analysing the possibilities of how Slovenia could efficiently contribute to the formation of such a system on the global stage will add to the body of scientific knowledge in international relations. Lastly, in the context of WP9 (“AI readiness indicators”), by creating a plan for the creation of a single information source that will offer an overview of the state of AI readiness in Slovenia in science, the economy, and public administration, contributions will be made to scientific knowledge in the field of computer science and informatics.
Furthermore, due to the fact that an International Research Centre for AI is to be founded in Slovenia and tasked with advising governments, organisations, legal entities, and the public on systemic and strategic solutions in implementing AI in various fields, the research conducted in this project will serve as a useful scientific basis for the scientific work that will ensue in the following years.
RELEVANCE TO THE DEVELOPMENT OF SLOVENIA
Competitiveness in the global market
Today, AI systems are able to reduce human efforts in a number of areas and companies thus employ them to optimise activities in industry. With the help of AI, business enterprises hope to reduce human efforts, obtain more accurate, efficient, and faster results, and minimise the danger of errors. If Slovenian companies wish to stay competitive in the global market it remains vital that existing and potential entrepreneurs understand the possibilities that AI brings as well as the harm that it may potentially inflict. By looking at how AI affects businesses, the public sector, and individuals, one of the goals of WP2 will be to try and discern the opportunities that are arising for business actors, aside from individuals and the public administration. In WP2 we will, inter alia, focus specifically on the following economic sectors: analytics in the banking sector, risky receivables and clients, analytics in the insurance sector, the analytics of insurance fraud (where we intend to collaborate with, e.g., the company Optilab), and analytics in business processes. AI is, for example, already widely used in banking and financial systems to invest money in stocks, carry out financial operations, manage various properties, etc. New technologies likewise affect heavy industries that are using AI to give shape to a specific object, to move objects from one place to another, as well as for company management in general, e.g. to store company data, to extract such data with ease, and to identify organisational gaps. In air transport, for instance, a large number of management processes are controlled by AI, while booking tickets as well as the operation of flights can be streamlined by means of AI. In entertainment, with the help of AI one can play video games with gaming bots and engage in virtual reality. Companies in various sectors will have to adapt, and the research conducted in the context of this project will assist them in recognising the possibilities of implementing trustworthy AI and avoid the negative effects of the new tech tools that are not ethically aligned with the values of society as enshrined in the constitutional legal framework. One of the positive effects of AI implementation that companies count on is that it will enhance the effectiveness and efficiency of their business and mapping the state of AI readiness in Slovenia, which we aim to realise in WP9 could assist companies in doing just that. The research done in WP2 and WP9 will allow companies in various fields to recognise the existing possibilities in Slovenia that they could take advantage of, and allow tech companies to recognise gaps that they could address and exploit as business opportunities.
Since the field of AI is still relatively new, legal regimes addressing AI issues are neither adequate nor coherent. One of the goals of this research will be to point to the possible legal clashes (WP4 and W7) that may occur in several separate legal disciplines as well as to question the suitability of the following legal regimes as regards the newly arisen needs of AI: competition law, privacy law, personal data protection law, consumer law, criminal and police law, and electoral law. Understanding what gaps exist in AI-related regulations and what challenges lie ahead is extremely important for existing companies as well as for those that are only starting a business in order for them to know what to pay attention to and where they might encounter some issues. By attempting to answer the question of what kind of ethics we have in mind when we speak of the ethics of AI, the research done in WP6 (“The ethical implications of AI”) will contribute to the body of knowledge regarding the regulation of AI, which likewise has important practical implications for businesses. Various ethical theories could serve as a value substructure for subsequent legal frameworks, from virtue ethics, deontological ethics, such as Kant’s, Pufendorf’s, or Lock’s ethics, and consequentialist ethics, top-down or bottom-up approaches, and so forth, and choosing one that leads to the most acceptable answers to the practical dilemmas of AI.
Changes in the labour market
The development of AI has serious implications for the future of work. McKinsey reports thatby the year 2030 between 75 million and 375 million people around the world may need to acquire new skills and change occupation, while in 60% of jobs, 30% of the activities that people do in that job could be automated (McKinsey, 2018). In WP4 we will devote special attention to discrimination and unequal treatment or mistakes that may occur as a consequence of AI use. An example thereof is reviewing job applications with AI, which can have negative consequences for job applicants. In WP5 we will address the question of how AI can be used to improve people’s psychological well-being, which is inevitably also relevant for an individual’s role as a worker since a massive amount of time today is spent at the workplace. Likewise, the mapping of AI use that we plan to undertake in WP2 will point to the professions that are most exposed to the changes that the new technologies bring, which will allow policymakers (WP3 and WP8) to plan ahead and mitigate possible harm in the form of unemployment.
Relevance for cultural development and protection of the natural and cultural heritage Furthermore, as the goal of WP2 is to map the various ways in which AI is used in Slovenia, Europe, and globally, the findings will also shine a light on whether AI could be used in cultural institutions as well as for the protection of the natural and cultural heritage. Similar conclusions can be drawn from the single information source, which will provide an overview of the state of AI readiness in Slovenia in science, the economy, and public administration, for which we intend to create a plan in the context of WP9.
Members of the research team
Inštitut za kriminologijo
Poljanski nasip 2