Artificial Intelligence Ethics for Well-being based on Sustainable Development Goals

Technology is a distinctive tool of the human species, aimed at improving survival perspectives and living conditions. It has played such a decisive role in human development that cultural evolution has increasingly taken precedence over biological changes: a Homo sapiens from 100,000 years ago would not be anatomically distinguishable from a modern human [135]. In the past century, the impact of technology has grown exponentially, with significant effects on both human lifestyles and the use of natural resources, as well as the balance of the Planet [218]. Since the 1970s, a clear awareness has emerged that uncontrolled human development could lead to catastrophic consequences for both the Planet and humankind. This marked the birth of ”Ecological Ethics” and the formulation of principles such as the ”Precautionary Principle” (Rio de Janeiro, 1992), and the ”Imperative of Responsibility” [149]. Greater attention began to focus on the impact of human activities on the environment. These early movements eventually grew into widespread global awareness, culminating in international conferences and reports such as ”Our Common Future” which laid the conceptual foundations for sustainable development. In the 1990s and 2000s, initiatives such as the ”Earth Summit” and the ”Millennium Development Goals” provided concrete and global momentum, ultimately leading to the creation of the Agenda 2030 for Sustainable Development in 2015, adopted by all United Nations member states. The Agenda 2030 introduced the 17 Sustainable Development Goals (SDGs), described as ”a shared blueprint for peace and prosperity for people and the planet, now and in the future”, which address: 1 - No Poverty, 2 - Zero Hunger, 3 - Good Health and Well-being, 4 - Quality Education, 5 - Gender Equality, 6 - CleanWater and Sanitation, 7-Affordable and Clean Energy, 8 - Decent Work and Economic Growth, 9 - Industry, Innovation, and Infrastructure, 10 - Reduced Inequalities, 11 - Sustainable Cities and Communities, 12 - Responsible Consumption and Production, 13 - Climate Action, 14 - Life Below Water, 15 - Life on Land, 16 - Peace, Justice, and Strong Institutions, and 17 - Partnerships for the Goals. The SDGs encapsulate global moral values for humanity in the 21st century, covering society, the economy, and the environment. For each of these, several targets have been established that can be regarded as ethical actions to be pursued in the short to medium term. The 2030 Agenda can be considered a strategic reference for the ethical development of technology. At the heart of the SDGs is the well-being of both humanity and the Planet. Throughout history, the concepts of well-being have played central roles in various ethical traditions. In the West, as advocated by the Stoics, Socrates, Plato, and Aristotle, and in the East, as exemplified by Zen and Buddhist teachings, these insights have been deeply valued. In recent years, wellbeing has gained increasing attention and is no longer solely focused on humanity but has expanded to include natural ecosystems and the entire Planet [72]. Well-being has become central to economic, political, social, and educational strategies aimed at ensuring the flourishing of present and future generations [45]. In recent decades, several influential theories have provided the theoretical foundations for developing well-being measurement tools and shaping a comprehensive assessment of human flourishing. One concrete approach to implementing these theories is through the metrics encapsulated within the SDGs [141]. Artificial Intelligence (AI) is the latest and most formidable human technology, with enormous potential to impact human development. It can contribute significantly to bringing benefits, but it also carries risks. It is no coincidence that some consider it a potential Existential and Catastrophic Risk [44] [17]. This underlines the urgent need to define and concretely apply an ethics of AI to guide its design, development and use. In this thesis, the ethical challenges of AI are analyzed, and an AI ethics framework is presented, aimed at supporting well-being and adopting the SDGs as ethical reference values for the flourishing of humanity and the Planet. Some scientific evidence is provided demonstrating that AI can support the achievement of the SDGs. This analysis is further supported by original research on Computer Vision (CV), a subfield of AI, where we an alyze over 8,000 articles to highlight its potential impact in achieving the targets of the Agenda 2030. We also examine potential conflicts between the SDGs related to human and planetary well-being. We introduce the 9 Planetary Boundaries (PBs) [218] as constraints for development that take into account the planetary ecosystem equilibrium and resources, to ensure the survival of the Planet and future generations. We show how CV can promote this type of sustainable development by supporting Vision 2050, which envisions ”a world in which more than 9 billion people can live well, within the planet’s boundaries, by 2050” [278, p. 4]. A number of concrete challenges in the practical implementation of AI ethics are presented, along with an illustration of four projects funded by the European University for Well-being (EUniWell) initiative, conducted in collaboration with a total of nine European universities, which highlight specific actions to address some of these challenges. These projects emphasize that it is essential to cultivate well-being within individuals for it to manifest outer, in society and in the world. This process involves the development of inner skills, both in terms of those that nourish well-being itself, and those that enable its realization [143]. Inner development begins with education supported by a culture of well-being, ongoing literacy, training, and continuous review. The four projects highlight the need to adopt work methodologies based on well-being, which facilitate the emergence of open, intuitive, collaborative, and effective collective intelligence within working groups [143]. The guiding principle is that, to bring about well-being and change the world, we must work with well-being. Chapter 2 introduces the concept of intelligence and AI, and presents the main AI technologies and architectures, with particular reference to Deep Neural Networks (DNNs). Chapter 3 synthesizes the AI strategies pursued by China, the US, the EU, and Italy. Chapter 4 presents the main European initiatives addressing ethical issues related to AI from 2019 to 2024, culminating in the approval of the EU AI Act. Chapter 5 discusses the ethics of well-being and proposes an AI ethics framework for well-being based on the SDGs. Chapter 6 analyzes the impact of AI and CV on achieving the SDGs, with a specific section dedicated to studying the impact of CV in realizing Vision 2050. Chapter 7 presents the four experimental projects investigating how to support the practical implementation of AI ethics. Finally, Chapter 8 presents the conclusions of this study.