by Georgia Kapitsaki (University of Cyprus) and Erwin Schoitsch (AIT Austrian Institute of Technology)
Software creation makes our daily lives easier, tackling different aspects of our daily activities (e.g., work, leisure, mobility, communication, smart homes, digital governance of public authorities, etc.). Applications are linked with several ethical issues and require ethically aligned design, with software as key factor, covering a wide range of domains, from healthcare to military systems. In order to fully benefit from any type of application, the alignment of software applications, embedded into systems and systems-of-systems, with ethical values and principles is of tremendous importance. Constructing software and systems in an ethical way is linked with many specific societal values, including individuality, equality, and democracy. Approaches and mechanisms that assist in producing ethics-aware software and systems are vital, in order to protect the main rights of the end-users and all other affected entities.
According to the Stack Overflow Developer Survey in 2018, “Developers are not sure how they would report ethical problems and have differing ideas about who ultimately is responsible for unethical code.”
Ethical aspects are becoming a key challenge particularly in Europe, based on European values with respect to human rights, human independence, freedom of decision and self-determination, democracy, and privacy. This results in the requirement raised in recent EC research projects (Horizon Europe, KDT Key Digital Technologies) to not only take care of standardisation, but also ethical and societal aspects more than before, by, for example, including an “Ethical Advisor” as a role or task. This was also the case for example, in recent Austrian national-funded projects; for example, the guest editor Erwin Schoitsch took over this role in a few European (AI4CSM, FOCETA, AI-IQ Ready) and national projects (ADEX – Autonomous Driving Examiner, Austrian Research Promotion Agency FFG/BMK).
Several informatics standardisation organisations, national Ethics Commissions, and UNESCO have published ethics guidelines, demonstrating the rising awareness on a global level, most of them focusing on AI and decision-making, but also on general issues like the requirement of developers and deployers to undertake an “Ethical Impact Assessment”.
IEEE in “Ethically Aligned Design” [L1] and UNESCO in “Recommendation on the Ethics of Artificial Intelligence” [L2] (see Figure 1) are referring to key values like “respect, protection and promotion of human rights and fundamental freedoms and human dignity”, “environment and ecosystem flourishing”, “ensuring diversity and inclusiveness”, and “living in peaceful, just and interconnected societies”. Key principles to follow are, among others, “proportionality and do no harm”, “safety & security, fairness and non-discrimination”, “sustainability, right to privacy and data protection”, “human oversight and determination”, “transparency and explainability, responsibility and accountability”, “awareness and literacy”, and “multi-stakeholder and adaptive governance and collaboration”. Applying ethical design in all steps of the software and systems engineering process is thus required nowadays, while different requirements may stem from the different application domains of software engineering (e.g., healthcare) and functions in system context.
Figure 1: IEEE Ethical aligned Design (2019) and UNESCO Recommendations (Nov. 2021).
The following paragraphs provide a short overview over the various contributions, grouping them according to their specific topics.
Artificial Intelligence ethics and challenges
Rottembourg uses as motivation the Digital Services Act and the Digital Markets Act to discuss relevant regulations in Artificial Intelligence and their consequences for auditing, focusing mainly on the entailed algorithmic challenges.
Biegelbauer et al. also stress out the importance of adapting the existing workflow of developing ethical Artificial Intelligence applications. In this respect, they are also presenting the role of the AIT AI Ethics Lab created at the AIT Austrian Institute of Technology.
Leikas et al. also focus on human-centricity and ethics in the development of Artificial Intelligence. They bring to our attention the Ethics Exercise Tool by the Finnish Centre for Artificial Intelligence that entails anticipation, reflexivity of one's own actions, inclusion, and responsiveness.
Artificial Intelligence and humans
Teixeira et al. present how machines and humans differ when performing classification tasks, and they study this specifically for classifying technological concepts and papers, in order to examine risks and vulnerabilities.
Héder stresses the need for the software and systems engineer of the 21st century to have knowledge of humanities and social sciences, and presents the Human-Centered AI Masters Programme launched this year in four countries towards this direction.
Ethical aspects of online data and web
Rossi focuses on the privacy and protection obligations of online data. The author presents the approach followed to gather examples of manipulative designs that can be found in online services and discusses the ethico-legal implications.
Lafon presents Ethical Web Principles that have been created by the Technical Architecture Group of W3C. The author explains the ethical aspects entailed when building web technologies, applications, and websites, reminding that human rights, dignity, and personal agency also need to be supported.
Noureddine et al. stress out the importance of reducing the energy and carbon footprint of software, and present a number of tools that assist software engineers in measuring and optimising the energy consumption of software systems.
Radersma discusses code optimisation for minimal energy consumption and gives practical recommendations that software engineers can follow in their everyday activities.
Dolas et al. address energy in the context of scientific computing ecosystems. To show how they are addressing this issue, they present as case study a strong scaling study on the computational fluid dynamics solver Palabos.
System design fairness and privacy
Valoggia presents the concept of a risk-based approach to engineering as a means to avoid the pitfalls of engineering to fail by not addressing properly the multidisciplinary characteristics of complex advanced technologies-based systems, especially in the context of Artificial Intelligence and Autonomous Systems.
Gornet et al. examine aspects that affect the design of a facial authentication system in the context of fairness, and discover aspects that concern data processing, neural networks used, and the training and evaluation of the systems.
Strobl et al. present an approach for data security by design in the implementation of a privacy-preserving biometric matching system. They describe three possible technical solutions: a reliable system, authentication through a distributed ledger, and use of homomorphic encryption methods.
Designing for sustainability and pandemics management
Ferrari et al. introduce requirements engineering research for sustainability and present how this can be achieved using interviews from 30 cross-disciplinary experts in the representative domain of digitalisation in forestry, agriculture, and rural areas.
Rainer et al. present the “ROADS to Health” project that focuses on measures and requirements for addressing cases of pandemics from different views in order to cover various needs. Ethical aspects and human needs are identified as key aspects in this context.
Note: A related topic “Machine Ethics” was discussed in the “Research and Society” section in ERCIM News 122, July 2020, p. 4 – 11.
[L1] Ethically Aligned Design, Version 2, IEEE, retrieved Oct. 2022: https://standards.ieee.org/wp-content/uploads/import/documents/other/ead_v2.pdf
[L2] Recommendation on the Ethics of Artificial Intelligence, UNESCO, 2021, retrieved Oct 2022: https://www.unesco.org/en/legal-affairs/recommendation-ethics-artificial-intelligence
University of Cyprus, Cyprus
AIT Austrian Institute of Technology, Austria