- Download 2
- File Size 1.10 MB
- File Count 1
- Create Date July 19, 2023
- Last Updated July 19, 2023
D5.1 Ethics State-of-the-Art and Literature Review
This deliverable provides a literature review of several areas of research and investigation in the field of ethics, broadly conceived, that are relevant to the topic areas of the GuestXR project. As the GuestXR project brings together research on virtual reality (VR) and VR systems with artificial intelligence (AI) and reinforcement machine learning techniques, as well as neuroscience and social psychology, the state-of-the-art provided in this deliverable includes discussion
of the following areas:
- the ethics of AI;
- ethical considerations related to the development and use of VRsystems and intelligent virtual environments (IVE), bringing ethical investigations of VR into dialogue with the ethics of social AI systems;
- value-sensitive design methodologies, to be iterated within the GuestXR project
The deliverable utilises a semi-systematic review approach. Semi-systematic reviews map dominant themes that have emerged over time, drawing attention to how topics have developed across different research traditions. This allows for the synthesis of multiple perspectives across several areas of research. Using this approach provides a broad overview of the main issues, and helps to make sense of complex and often contested topics. A semi-structured review can then be used as a decision aid. This review will serve as the knowledge basis for the development of an “ethics-by-design” approach that will be utilised throughout the
In relation to the development of AI and machine learning applications, this review loosely adopts a framework proposed by Carly Kind, director of the Ada Lovelace Institute. This framework looks at the development of the rapidly expanding field of AI ethics in three, now overlapping phases:
- Abstract guiding principles and norms generally derived from systematic approaches in philosophical ethics and applied ethics;
- More technically oriented approaches that look primarily at how interventions and work programmes led by computer scientists and developers can address questions concerning issues like fairness, bias, and accessibility. This approach operationalised many of the more abstract or high-level ethical concerns raised in the first phase, but also tended to reframe them as technical problems requiring technical solutions.
- A third phase has shifted the discussion of ethical AI towards more social and political questions relating to questions of justice, including “social justice, racial justice, economic justice, and environmental justice” (Kind, 2020). Within this phase, technologies are considered embedded in and co-producing socio-technical systems. As such, it is also concerned with questions relating power and structure as well as highlevel ethical concerns, legal constraints and technical “fixes”. The practical focus in this approach is on co-creation processes that involve affected individuals and communities at early stages of technological development and innovation processes, giving them voice and capacity for action within the design and application of technical systems.
The third phase is also characterised by the further development of Responsible Innovation and “by-design” approaches. These approaches also characterise the approach to ethics within the GuestXR project.
The European Commission’s Ethical Guidelines for Trustworthy AI and the related seven requirements produced by the High-Level Expert Group on Artificial Intelligence (AI HLEG) nonetheless remain extremely important foundational guidelines for the development of ethics-by-design methodologies and practices within the GuestXR project.
Based on fundamental rights and ethical principles, the Guidelines list seven key requirements that AI systems should meet in order to be trustworthy:
- Human agency and oversight
- Technical robustness and safety
- Privacy and Data governance
- Diversity, non-discrimination and fairness
- Societal and environmental well-being
These requirements are based on the fundamental rights elaborated in the European treaties and other foundational documents as well as foundational ethical principles.
Ethical and social consideration of VR systems are closely intertwined with technical, scientific and indeed philosophical questions about embodiment, immersion and presence within VR environments. Thus, ethical examination of VR systems and environments has evolved alongside technical, psychological and sociological investigations. An initial focus has been on potential harms to individuals or vulnerable groups in VR environments – and how to prevent them. This focus has addressed questions concerning discrimination, stereotyping, anti-social behaviour such as sexual harassment as well as accessibility relating to equipment and the affordances within VR environments.
As the potential use of VR environments in various arenas of social life has increased and with many companies now eyeing the commercial potentials of these technologies, the nascent field of VR ethics has followed the development of AI ethics in starting to increasingly take into consideration broader sociopolitical
questions relating to the governance of immersive virtual environments.
As the possibilities for individuals to spend greater amounts of time within VR environments (for professional, leisure, and commercial activities) continues to increase, the field of VR ethics has also begun to explore how habits and affinities developed within VR environments might transfer into physical reality. This has opened up further possibilities for the use of VR, such as in therapeutic contexts. The growing potential of VR and the expanding scope of use applications has made the necessity for serious ethical consideration and upstream value-sensitive intervention increasingly important.
The GuestXR project, for this reason, has committed to the implementation of ethics-by-design approaches within the scientific work programme of the project. This deliverable will be followed by further guidelines for implementation of ethics-by-design methods and processes. These guidelines will therefore build, at least in part, upon the initial review of ethical and related literature in this deliverable.