Principled Artificial Intelligence: Mapping Consensus in Ethical and Rights-Based Approaches to Principles for AI

“The HCAI movement has raised these issues again with frequent calls for Algo-rithmic Accountability and Transparency [40], Ethically Aligned Design [56], and Professional Responsibility [36].”

“…risks could potentially be ameliorated through the application of principles defining “ ethical AI ” , “ responsible AI ” or “ AI for Good ” that have emerged in the last years [115,116], especially those that address fairness, non-discrimination, accountability, transparency, privacy and security.”

“…derived from a review of six high-profile initiatives, bears substantial similarities with other scholarly efforts to AI create ethics typologies (Fjeld et al. 2020; Jobin et al. 2019; Zeng et al. 2018), and is highly familiar given its overlap with traditional bioethics principles (Beauchamp…”

“Generally speaking, accountability in AI relates to the expectation that designers, developers, and deployers will comply with standards and legislation to ensure the proper functioning of AIs during their life cycle (Fjeld et al., 2020).”

“The expansion of interest and application of AI research has triggered widespread public and government scrutiny of the ethical and responsible principles to guide designers, managers, and policy makers (Fjeld et al., 2020; IEEE, 2019).”

“It has a strong emphasis on ethical guidelines and principles (Fjeld et al., 2020; Jobin et al., 2019) which includes a number of high profile interventions (AI HLEG, 2019) but it also legislation such as the EU ’ s AI Act (European Commission, 2021), various approaches to standardisation (IEEE,…”

“Such adverse outcomes of AI use, along with the threat of loss of control or autonomy over superior AI entities, has sparked an ongoing debate about the need to establish a set of principles to effectively govern AI (Barredo Arrieta et al., 2020; Fjeld et al., 2020).”

“They propose solutions such as differential privacy techniques to protect individual privacy while ensuring transparency (Fjeld et al., 2020).”

“…Trustworthy AI of the High-Level Expert Group on AI set up by the European Commission (High-Level Expert Group on Artificial Intelligence, 2019), and the AI4People’s principles for AI ethics (Floridi et al., 2018; see Fjeld et al., 2020; Hagendorff, 2020; or Jobin et al., 2019 for a meta-analysis).”

“Part of this increasing gap between laws and technology is caused by the lack of a thorough and accurate definition of AI (Firth-Butterfield, 2017; Larsson, 2020), which is aggravated by the fact that the definition changes as the technology evolves (Fjeld et al., 2020).”

“As a last mention, we would like to cite the work done by Fjeld et al., 20 one of the first to present documents from Latin America.”

“In particular, the definition draws on that of AI governance for business [12] while incorporating the regulatory constraints [19, 31] and ethical AI principles [6, 32].”

“…Institute of Standards and Technology (NIST) 6 , International Organization for Standardization (ISO) 7 , and professional bodies e.g., the Institute of Electrical and Electronics Engineers [IEEE] [9] have published their ethical AI principles and guidelines [7, 10] for the governance of AI.”

“These documents and reports were surveyed, sometimes with analytical and critical insights, by some recently published papers [291, 292, 249, 293, 206].”

“This paper sets out to review the landscape of suggested ethical frameworks with a focus on those which go beyond high-level statements of principles (see [4–6] for review of principles), and offer practical tools for application of these principles in the production and deployment of systems.”

“AI-based decision-making systems can be used for scalable and larger ecosystems; however, the systems need to be incorporated with some principles related to ethics and rights (Fjeld et al., 2020).”

“…of such systems: bias amplification, data privacy, lack of transparency, human oversight, accountability, etc. (Royal Society (Great Britain), 2017; Fjeld et al., 2020; Campolo et al., 2017; Floridi et al., 2018; Floridi, 2019, 2021; High-level expert group on artificial intelligence (HLEG, AI),…”

“A similar pattern has been observed in AI ethics, as recent statements of principles share many common elements (Fjeld et al., 2020; Jobin et al., 2019).”

“2017), ethics frameworks (AI HLEG 2019), development methodologies (Martin and Makoundou 2017; Fjeld et al. 2020) etc.”

“envisioning documents [38, 43, 59], the EAI guidelines present a high-level thematic checklist for anyone engaged with “data about”

“It provided the authors with a list of ethical principles specific to the field of AI (Hagendorff 2019; Jobin et al. 2019; Fjeld et al. 2020).”

“Finally, there is the ethical perspective (Whittle, 2019; Fjeld et al., 2020).”

“[8] 2020 study of a different sample of documents shows, while there are certainly points of convergence, by no means is there unanimity across the 36 principle documents reviewed.”

“According to Fjeld et al. (2020), transparency is the assertion that AI systems should be designed and implemented in such a way that oversight is possible.”

“A sizable amount of prior work views socially responsible or beneficent AI as a consensus value [4], [6].”

“As found in other, smaller, AI literature collections and synthesis papers, the proliferation of AI documents is coming from the Northern or Western world [13, 14, 86, 87].”

“Fjeld et al. (2020) conducted a review of 36 key documents and discovered that they all adhered to fairness and non-discrimination principles.”

“Previous scoping reviews on AI ethics guidelines (Fjeld et al., 2020; Hagendorff, 2020; Jobin et al., 2019) congruently found a set of reoccurring paramount principles for AI development and use: transparency, fairness, security, safety, accountability, privacy, and beneficence.”

“Drawing upon the combined bibliographies of three highly cited previous reviews of AI ethics guidelines [5, 11, 12] we conducted a review of the AI ethics literature that would closely observe how AI business practices, their political and economic effects, and the ethical dimensions of that…”

“Increasing concern over bias in automated systems has led to an outcry for companies incorporate fairer and more transparent systems [20].”

“…from Harvard University’s Berkman Klein Center for Internet & Society is a survey of AI principles documents that have been published around the globe in recent years [29], including the OECD Principles on Artificial Intelligence that the United States and 41 other nations adopted in 2019 [30].”

“As said before, the four basic AI virtues cover all common principles of AI ethics as described in prior discourses (Fjeld et al., 2020; Floridi et al., 2018; Hagendorff, 2020; Jobin et al., 2019; Morley et al., 2020).”

“In addition to the principles of autonomy, non-401 maleficence, beneficence, and justice, recurring themes of AI ethics are for example transparency, explainability, accountability, and fairness [37, 38].”

“In an analysis by Fjeld et al. ( 2020 ) of AI Ethics Principle-type documents, over 81% contained some recommendations in favour of ‘safety’ and ‘security’.”

“…and family life” (Council of Europe, 2020) Users should be allowed to determine if and how their personal data is collected, processed and disseminated (Fjeld et al., 2020, p. 21; IEEE, 2021, p. 71) GDPR Data Protection Impact Assessment (European Commission, 2016) Controls to limit data use.”

“…of frameworks (Jobin et al., 2019; Schiff et al., 2020), the differences between the principles included, their hierarchies and recurrences (Fjeld et al., 2020) and the way in which these principles are defined, represent an epistemic and operational challenge for the various AI…”

“The principle of the promotion of human values is largely congruous with fundamental human rights, and includes the following three main concerns: supporting and promoting human values and human flourishing, benefiting society, and ensuring broad access to technology (Fjeld et al., 2020).”

“Many responsible AI interventions orient towards developer-facing interventions, such as developing AI principles [57], educating practitioners to foster ethical awareness [93], and moderating generative AI systems through training data mitigations, in-model controls (e.”

“Scholars have identified different high-level ethical principles that should govern the development of AI systems [25, 48, 97].”

“Most AI regulatory documents and guidance are based on common principles [13], which are referred to as “responsible AI.””

“As mentioned in the introduction, at least three meta-studies on AI ethics guidelines [26, 35, 36] confirm that explainability, fairness, privacy, safety, and accountability are mentioned in almost every code of ethics.”

“Such tools and methodologies can draw on and be inspired by a large number of ethics principles and frameworks 45,46 and find their expression in ethics codes 47 which may be part of professional guidance 48 and include process of standardisation 49,50 and certification 51,52 .”

“…through an increase in academic publications on the topic (For example the Berkman Klein Center identified 36 “prominent AI principles” documents [16]), the involvement of AI ethics in the policy forum (e.g., European Commission High-Level Expert Group on AI) [17], and the adoption of AI ethics…”

“Out of the numerous ethics frameworks and guidelines produced by governments and organisations, a broad consensus is emerging around what the major principles of AI ethics should be [9].”

“The first principle, promotion of human values, suggests that AI-based systems should be used for the common good and should be developed consistently with cross-cultural human values [8].”

“…documents published by civil society, the private sector, governments, intergovernmental organizations, and multi-stakeholder organizations by (Fjeld et al., 2020).14 Wachter et al. (2017) is an excep- 13They do list several texts that focus on the autonomy of those who deploy the algorithms,…”

“of values like fairness, transparency, and accountability in building AI [11], for several reasons.”

“…as it mainly emerges, to maintain people’s full control over themselves and AI systems (Floridi et al. 2018; IEEE 2019; HLEG-AI 2019; OECD 2019; Fjeld et al. 2020), as well as to privacy-preserving techniques based on informed consent (European Parliament 2017; IEEE 2017; Google AI 2018; WHO…”

“This is why it was soon derided in a letter signed by 156 experts on AI, ethics, and law [31].”

“54 In the context of AI in health-care, this means involving patients and carers in the design and deployment of AI tools.”

“Over the past few years, corporations, governments, civil society groups, and multi-stakeholder organizations have published dozens of high-level AI ethics principles (Fjeld et al., 2020).”

“This is somewhat surprising since concerns about the “ explainability ” and “ transparency ” of AI systems have long dominated research and policy-making conversations around what makes for an ethical AI (cf. Fjeld et al., 2020; Jobin et al., 2019).”

“For instance, studies have been carried out to analyse ethical principles of AI [10, 21, 25, 42].”

“Case studies on implementing HAI guidelines reveal stakeholders’ struggle with concepts of privacy and fairness [38, 60, 129].”

“Studies comparing existing guidelines found that they are converging towards the same set of principles, even more in recent times (Jobin et al. 2019; Fjeld et al. 2020).”

“As Gunkel (2012) and Fjeld (2020), have demonstrated, the strength of the Floridi framework does not lie in its normative rigid - ity; rather, it lies in its capacity to engage in dialogue with other approaches and to serve as a basis for ethical evaluation in scenarios where specific regulations…”

“Studies that have dealt with responsible AI frameworks emphasize the importance of high-level principles such as transparency, justice, fairness, and nonmaleficence (Fjeld et al., 2020; Hagendorff, 2020; Jobin et al., 2019).”

“, who found that autonomy typically provides the theoretical grounding for principles of ‘human control of technology’ [14].”

“Inspired by the research of Morley et al. (2020), Fjeld et al. (2020), and Ayling and Chapman (2021), we created a diagram found at Tools 4Resp onsib leAI, specifically in the ‘Typology’ section of the website.”

“…AI ethics focusing on the eight global themes (“privacy,” “accountability,” “safety and security,” “transparency and explainability,” “fairness and non-discrimination,” “human control of technology,” “professional responsibility,” and “promotion of human values”) developed by Fjeld et al. (2020).”

“The growing societal implications of AI have prompted extensive attention to the ethics principles that should guide AI development and use [1, 2] and, more recently, some attention to how organizations can translate these principles into practice [3, 4, 5].”

“We use the ethical principles listed in Harvard University’s mapping study [2].”

“However, limited, if any, research has been done on ascertaining the current AI regulatory landscape in the Global South, and in Africa, in particular (Wang and Siau, 2018; Jobin et al., 2019; Fjeld et al., 2020; Gerke et al., 2020).”

“” There have been many critiques of “principles” based approaches to AI Ethics and AI Governance, yet organizations seemingly turn to this as a first step and too often conclude that this is the only step toward what is needed to be done (Jobin et al., 2019; Whittlestone et al., 2019; Fjeld et al., 2020).”