Table 1. AI Ethics principles in... | Emerald Publishing

Table 1.

AI Ethics principles in computer science (Kieslich et al., 2022, p. 6; Jobin et al., 2019)

AI ethics principles	Descriptions
Explainability	Explainability enhances transparency and acts as a key indicator for measuring it. By clarifying processes, explainability helps minimize harm and improve both AI tools and trust. For instance, being transparent involves explaining how data is sourced and used, as well as how automated decisions are made
Fairness	Fairness is closely linked to equity and justice, focusing on preventing, monitoring, and mitigating unwanted bias and discrimination. This involves acquiring and processing data, especially training data, that is accurate, complete and diverse
Security	Security, in relation to safety, serves as a key indicator for evaluating “non-maleficence.” the use of AI should not cause harm, such as discrimination, privacy violations, or physical, psychological, or emotional harm and should also consider socio-economic aspects such as social well-being and infrastructure
Accountability	Accountability measures responsibility, requiring individuals to act with integrity and comply with legal obligations
Accuracy	Accuracy means accepting almost no errors and can be used to evaluate societal well-being
Privacy	Privacy is not only a value to uphold but also a fundamental right to be protected, often involving the safeguarding of data and ensuring data protection
Machine autonomy	A computer system is capable of functioning properly within a defined scope without human supervision

AI ethics principles	Descriptions
Explainability	Explainability enhances transparency and acts as a key indicator for measuring it. By clarifying processes, explainability helps minimize harm and improve both AI tools and trust. For instance, being transparent involves explaining how data is sourced and used, as well as how automated decisions are made
Fairness	Fairness is closely linked to equity and justice, focusing on preventing, monitoring, and mitigating unwanted bias and discrimination. This involves acquiring and processing data, especially training data, that is accurate, complete and diverse
Security	Security, in relation to safety, serves as a key indicator for evaluating “non-maleficence.” the use of AI should not cause harm, such as discrimination, privacy violations, or physical, psychological, or emotional harm and should also consider socio-economic aspects such as social well-being and infrastructure
Accountability	Accountability measures responsibility, requiring individuals to act with integrity and comply with legal obligations
Accuracy	Accuracy means accepting almost no errors and can be used to evaluate societal well-being
Privacy	Privacy is not only a value to uphold but also a fundamental right to be protected, often involving the safeguarding of data and ensuring data protection
Machine autonomy	A computer system is capable of functioning properly within a defined scope without human supervision

Source(s): Table created by authors and adapted from Kieslich et al. (2022, p. 6) and Jobin et al. (2019)