CoChat
Sign Up
Cited by 269Ethics and Social Impacts of AIEthics in Business and EducationFlooding and Environmental Impact

A high-level overview of AI ethics

Open access
Kazim, Emre, Koshiyama, Adriano Soares·2021·Patterns
DOI 10.1016/j.patter.2021.100314PMC PMC8441585
Open this paper in CoChat — free accountOpen PDF

Abstract

Artificial intelligence (AI) ethics is a field that has emerged as a response to the growing concern regarding the impact of AI. It can be read as a nascent field and as a subset of the wider field of digital ethics, which addresses concerns raised by the development and deployment of new digital technologies, such as AI, big data analytics, and blockchain technologies. The principle aim of this article is to provide a high-level conceptual discussion of the field by way of introducing basic concepts and sketching approaches and central themes in AI ethics. The first part introduces concepts by noting what is being referred to by "AI" and "ethics", etc.; the second part explores some predecessors to AI ethics, namely engineering ethics, philosophy of technology, and science and technology studies; the third part discusses three current approaches to AI ethics namely, principles, processes, and ethical consciousness; and finally, the fourth part discusses central themes in translating ethics in to engineering practice. We conclude by summarizing and noting the inherent interdisciplinary future directions and debates in AI ethics.

AI snapshot

Written from the paper’s full text.

TL;DR

This article offers a high-level conceptual overview of AI ethics as a nascent, interdisciplinary subfield of digital ethics, tracing its predecessors and current approaches. It argues that AI ethics spans principles, processes, and ethical consciousness, and identifies themes for translating ethics into engineering practice.

Key findings

  • AI ethics is framed as a subset of digital ethics, covering the psychological, social, and political impacts of AI; it is described as highly interdisciplinary and rapidly evolving.
  • The paper sketches three current approaches to AI ethics: principles, processes (including ethical-by-design and governance), and ethical consciousness.
  • The principles approach is said to be plagued by vagueness, incongruence, unclear terminology, and lack of consensus; as of April 2020, there were over 80 AI statements of principles.
  • A 2019 review cited by the paper identified eleven “ethical principles” in AI ethics guidelines, listed in order of prevalence: transparency; justice, fairness, equity; non-maleficence; responsibility and accountability; privacy; beneficence; freedom and autonomy; trust; dignity; sustainability; and solidarity.

Methods

  • Design: a high-level conceptual overview and introduction to AI ethics, structured into four parts: concepts and terms; predecessors; three current approaches; and themes in translating ethics into engineering practice.
  • Data or sample: no empirical dataset or sample is used; the article draws on literature from the humanities and engineering.
  • Main analysis: conceptual synthesis and high-level analysis of definitions, predecessor disciplines, approaches, and major themes, rather than systematic review or empirical testing.

Limitations

  • The paper states that its concepts are “not to be read as a series of definitive definitions” but as introductions and points of departure, and that readers should engage critically with them.
  • The paper states that its selection of three predecessor disciplines is not exhaustive and notes that other bodies of literature, perhaps most importantly the ethics of robotics, are not discussed.
  • The paper states that the predominant discourse of AI ethics is found within Western European and North American contexts, which it also reads in terms of access to English-language literature; evident from the text, the overview is not a systematic review and offers no empirical validation of its claims.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/j.patter.2021.100314

Cites

73 of 73 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 42 cited sources checked · 0 supported · 0 flagged

  • Findingcitation

    AI principles are mostly vague and thereby difficult to interpret.

    “First, the principles are mostly vague and thereby difficult to interpret (e.g., the accountability principle “individuals and organizations need to be held to account,” does not provide an expansive explanation of how this would look in practice).52,53”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [52] Verma S. Rubin J. Fairness definitions explained 2018 IEEE/ACM International Workshop on Software Fairness (FairWare) 2018 IEEE 1 7

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [53] Dwork C. Hardt M. Pitassi T. Reingold O. Zemel R.S. Fairness through awareness arXiv 2011 CoRR abs/1104.3913 (2011). preprint arXiv:1104.3913 source

      Not checked in the free run.

  • Findingcitation

    AI principles are incongruent, i.e., within the same set of principles, the individual principles contradict one another.

    “Second, the principles are incongruent, i.e., within the same set of principles, the individual principles contradict one another (e.g., both the principle of “openness” and the principle of “privacy” are asserted together, where respect for one is likely to come at the cost of implementation and respect for the other).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [52] Verma S. Rubin J. Fairness definitions explained 2018 IEEE/ACM International Workshop on Software Fairness (FairWare) 2018 IEEE 1 7

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [53] Dwork C. Hardt M. Pitassi T. Reingold O. Zemel R.S. Fairness through awareness arXiv 2011 CoRR abs/1104.3913 (2011). preprint arXiv:1104.3913 source

      Not checked in the free run.

  • Findingcitation

    There is a clear lack of consensus between various statements of AI principles.

    “Finally, although there is overlap between the various statements of principles, there is also a clear lack of consensus.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [5] Jobin A. Ienca M. Vayena E. The global landscape of AI ethics guidelines Nat. Mach. Intell. 1 2019 389 399 source

      Not checked in the free run.

  • Findingcitation

    Data Protection Impact Assessments (DPIA) are the most well-established form of auditing and impact assessment.

    “The most well-established form of this is Data Protection Impact Assessments (DPIA)69,70”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [69] Guide to the General Data Protection Regulation (GDPR) Information Commissioner’s Office https://ico.org.uk/media/for-organisations/guide-to-data-protection/guide-to-the-general-data-protection-regulation-gdpr-1-0.pdf 2019

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [70] Kazim E. Koshiyama A. The interrelation between data and AI ethics in the context of impact assessments AI Ethics 2020 10.1007/s43681-020-00029-w source

      Not checked in the free run.

  • Findingcitation

    There is considerable lack of clarity regarding terminology/concepts in AI principles.

    “Third, there is considerable lack of clarity regarding terminology/concepts (e.g., principles of accountability, transparency, and openness are complimentary and, in some respects, synonymous expressions).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [52] Verma S. Rubin J. Fairness definitions explained 2018 IEEE/ACM International Workshop on Software Fairness (FairWare) 2018 IEEE 1 7

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [53] Dwork C. Hardt M. Pitassi T. Reingold O. Zemel R.S. Fairness through awareness arXiv 2011 CoRR abs/1104.3913 (2011). preprint arXiv:1104.3913 source

      Not checked in the free run.

  • Classified as probably not a substantial claim. 63
  • May combine two claims. 26

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingcitation

    A comprehensive 2019 review of AI ethics guidelines found eleven ethical principles, identified through frequency of terms and listed in order of prevalence.

    “For example, a comprehensive 2019 review of AI ethics guidelines found eleven “ethical principles” namely (1) transparency; (2) justice, fairness, equity; (3) non-maleficence; (4) responsibility and accountability; (5) privacy; (6) beneficence; (7) freedom and autonomy; (8) trust; (9) dignity; (10) sustainability; and (11) solidarity. These principles were identified through frequency of the terms”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingcitation

    To date (April 2020) there are over 80 AI statements of principles.

    “Indeed, to date (April 2020) there are over 80 AI statements of principles.5”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [5] Jobin A. Ienca M. Vayena E. The global landscape of AI ethics guidelines Nat. Mach. Intell. 1 2019 389 399 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingcitation

    The AI ethics literature has significantly increased in the past number of years (2017-) and continues to evolve.

    “There is a growing and evolving literature that has significantly increased in the past number of years (2017–) and this literature continues to evolve with reviews from a computer science perspective5 and from the philosophical and humanities perspective.6, 7, 8”
    Quote in the paperNot a substantial claim0 of 4 sources checked
    • cites [5] Jobin A. Ienca M. Vayena E. The global landscape of AI ethics guidelines Nat. Mach. Intell. 1 2019 389 399 source

      Not checked in the free run.

    • cites [6] Müller V.C. Ethics of artificial intelligence and robotics Zalta E.N. The Stanford Encyclopedia of Philosophy 2020 https://plato.stanford.edu/archives/win2020/entries/ethics-ai/

      Not checked in the free run.

    • cites [7] Coeckelbergh M. AI Ethics 2020 MIT Press

      Not checked in the free run.

    • cites [8] Gordon J.-S. Nyholm S. Ethics of Artificial Intelligence 2021 Internet Encyclopedia of Philosophy

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Findingnone given

    There is considerable ambiguity and lack of consensus in the field of digital regulation and standards, making it difficult to establish best practices for translating principles into engineering practice.

    “As noted above, there is considerable ambiguity and lack of consensus in the field of digital regulation (perhaps with the exception of GDPR) and standards, thus making it difficult to establish best practices in the domain of translating principles into engineering practice.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingcitation

    There have been an increasing number of high-profile cases of harm resulting either from misuse of technology or from technology having design flaws.

    “Indeed, there have been an increasing number of high-profile cases of harm that has resulted either because of the misuse of the technology (e.g., psychometric voter manipulation,1,2 facial recognition surveillance, mass data collection without consent, etc.), or as a result of the technology having design flaws (e.g., bias in cases of recidivism,3 loan rejection, and medical misdiagnosis, etc.).4”
    Quote in the paperNot a substantial claim0 of 4 sources checked
    • cites [1] Krotzek J.L. Inside the voter’s mind: the effect of psychometric microtargeting on feelings toward and propensity to vote for a candidate Int. J. Commun. 13 2019 3609 3629

      Not checked in the free run.

    • cites [2] Costa E. Halpern D. The behavioural science of online harm and manipulation, and what to do about it: an exploratory paper to spark ideas and debate The Behavioural Insights Team Report 2019 1 82

      Not checked in the free run.

    • cites [3] Chouldechova A. Fair prediction with disparate impact: a study of bias in recidivism prediction instruments arXiv 2016 1 6 28632438

      Not checked in the free run.

    • cites [4] Bellamy R.K. Dey K. Hind M. Hoffman S.C. Houde S. Kannan K. AI Fairness 360: an extensible toolkit for detecting, understanding, and mitigating unwanted algorithmic bias arXiv 2018 1810.01943

      Not checked in the free run.

  • May combine two claims.

    Findingcitation

    More general "data ethics" canvases and process frameworks have emerged, and impact assessments specific to AI are being called for and developed.

    “To date more general “data ethics” canvases and process frameworks have emerged,71 and impact assessments specific to AI are being called for and developed.54,72”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [71] Data Ethics Canvas Open data institute https://theodi.org/article/data-ethics-canvas/ 2020

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [54] White Paper on Artificial Intelligence A European Approach to Excellence and Trust 2020 European Commission

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [72] Jordan S. Fazelpour S. Koshiyama A. Kueper J. DeChant C. Leong B. Creating a Tool to Reproducibly Estimate the Ethical Impact of Artificial Intelligence 2019 AI PULSE Papers

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingnone given

    There are debates concerning the value derived from personal data and the distribution of financial benefits derived thereof.

    “Furthermore, there are debates concerning the value derived from personal data and the distribution of financial benefits derived thereof (e.g., mass data-driven business models).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    Data stewardship has come to be a discipline and set of skills in its own right.

    “Stewardship is the management of this multi-layered process and has come to be a discipline and set of skills in its own right.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Findingnone given

    Privacy has emerged as a political concern, with mass surveillance and personal data being used to target and engage in recommendation and manipulation for both political and economic ends.

    “In addition, privacy has emerged as a political concern, with mass surveillance and personal data being used to target and engage in recommendation and manipulation (for both political and economic ends).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    Continental law, exemplified by the EU's legislative agenda and typified through a top-down first-principles legal philosophy, is less likely to have the problem of ambiguity not translating to automated compliance.

    “Regarding this specific concern, continental law (exemplified by EU’s legislative agenda) typified through a top-down (first-principles) legal philosophy is less likely to have this problem.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

    Premisereasoning

    Ethical consciousness encompasses the integration of codes of conduct and compliance, and expands to consider reputational issues, corporate social responsibility, and institutional philosophy and culture.

    “Sharing many of the themes from the previous section, it encompasses the integration of codes of conduct and compliance; however, it also expands to consider reputational issues, (corporate) social responsibility, and, most relevant to the development of ethical consciousness, concerns for institutional philosophy and culture.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationcitation

    The terms used in AI ethics guidelines overlap and require considerable disambiguation.

    “However, as noted above, there is overlap and these terms require considerable disambiguation.5”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [5] Jobin A. Ienca M. Vayena E. The global landscape of AI ethics guidelines Nat. Mach. Intell. 1 2019 389 399 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    In the common law tradition, based on case and precedent, the statutes accommodate ambiguity and contradiction.

    “Fourth, in the common law tradition (typically found in the anglophone), based on case and precedent, the statutes accommodate ambiguity and contradiction (relying”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisereasoning

    The two main approaches to AI ethics have been a principles approach and an ethical-by-design approach.

    “As a result, the two main approaches to AI ethics has been a principles approach, which can be read broadly as an attempt to guide and structure the uses of the technology (thereby mitigating the risk of misuse) and an ethical-by-design approach, which seeks to mitigate the harms that result from design flaws.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Human-centric AI is the emerging overarching value framework of AI ethics.

    “This section closes with an exploration of “human centric AI,” which we take to be the emerging overarching value framework of AI ethics.9”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [9] Lukowicz P. The challenge of human centric AI Digitale Welt 3 2019 9 10 source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    The ambiguity and contradiction accommodated by common law do not translate to automated regulatory compliance via expressing laws in codes or protocols.

    “This is a problem with respect to automated systems, where increasingly there is a call to automate regulatory compliance via expressing laws in codes/protocols (the ambiguity and contradiction accommodated by common law does not translate in this context).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Bio/medical ethics is well established, robust, has accountability mechanisms, and is an example of the ethics of an applied science with significant social impact.

    “Bio/medical ethics can be turned to for inspiration when approaching and developing AI ethics due to the fact that it is well established, robust, has accountability mechanisms and is an example of the ethics of an applied science with significant social impact60, 61, 62, 63, 64”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites [60] Mittelstadt B. Floridi L. The ethics of big data: current and foreseeable issues in biomedical contexts Sci. Eng. Ethics 22 2016 303 341 26002496 source

      Not checked in the free run.

    • cites [61] Mittelstadt B. Designing the health-related internet of things: ethical principles and guidelines Information 8 2017 77

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [62] Christine H. Making the ethical practical: ideas for applying a digital health ethical framework for charities. DataKind.org https://www.datakind.org/blog/making-the-ethical-practical-ideas-for-applying-a-digital-health-ethical-framework-for-charities 2019

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationcitation

    AI ethics is highly interdisciplinary and is rapidly evolving.

    “Like digital ethics, AI ethics is thus highly interdisciplinary and is rapidly evolving.6,10,11”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [6] Müller V.C. Ethics of artificial intelligence and robotics Zalta E.N. The Stanford Encyclopedia of Philosophy 2020 https://plato.stanford.edu/archives/win2020/entries/ethics-ai/

      Not checked in the free run.

    • cites [10] Kriebitz A. Lütge C. Artificial intelligence and human rights: a business ethical assessment Bus. Hum. Rights J. 5 2020 84 104 source

      Not checked in the free run.

    • cites [11] Floridi L. The 4th Revolution 2014 Oxford University Press

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given

    Engineering ethics is predominantly driven by self-assembled communities and associations, who develop their own standards, are process orientated, and typically go beyond legal compliance.

    “It is noteworthy that engineering ethics is predominantly driven by self-assembled communities and associations, who develop their own standards, are process orientated (e.g., case study exploration, etc.), and typically go beyond legal compliance.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisenone given

    AI ethics is a field that has emerged as a response to the growing concern regarding the impact of AI.

    “Artificial intelligence (AI) ethics is a field that has emerged as a response to the growing concern regarding the impact of AI.”
    • Also stated in: § Introduction · plain
      “Artificial intelligence (AI) ethics is a field that has emerged as a response to the growing concern regarding the impact of AI.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Interpretationreasoning

    From a design perspective, implementation of ethical principles will need to be balanced and traded-off.

    “Finally, and as a corollary to the previous point, from a design perspective, implementation of ethical principles will need to be balanced and “traded-off.””
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    The philosophy of technology literature is typically negative, highlighting dangers, risks, and loss of meaning through adoption of new technologies and increased technocratization.

    “The literature is typically negative, highlighting the dangers, risks, and loss of meaning through adoption of new technologies and increased technocratization with respect to the ordering principles of society.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given

    Legal compliance is a clear and objective standard by which to judge and evaluate ethics, where lawfulness is a necessary but insufficient condition of ethics.

    “Indeed, legal compliance is a clear and objective standard by which to judge and evaluate ethics (where lawfulness can be read as a necessary but insufficient condition of ethics).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Science and technology studies emerged circa the 1980s.

    “This empirical turn gave birth to science and technologies studies (circa 1980s–).43”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [43] Franssen M. Vermaas P.E. Kroes P. Meijers A.W.M. Philosophy of Technology after the Empirical Turn 2016 Springer

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationnone given

    The three approaches-theoretical and abstract, practical and process, and culture and society-are all necessary in the development of a mature AI ethics.

    “The above approaches can be thought of in terms of (1) the theoretical and abstract, (2) the practical and process, and (3) culture and society, and in turn can be thought of as all necessary in the development of a mature AI ethics, i.e., one that is truly reflective of the nuances and the inherent complexity found in all forms of applied ethics.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

    Interpretationnone given

    In the context of new digital technologies such as AI, education and training will require continuous updating.

    “in the context of new digital technologies, such as AI, education and training will require continuous updating”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Ethical consciousness can be read as coming out of business ethics, which is an applied ethics within the commercial environment.

    “Ethical consciousness can be read as coming out of business ethics,10 which is an applied ethics within the commercial environment.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [10] Kriebitz A. Lütge C. Artificial intelligence and human rights: a business ethical assessment Bus. Hum. Rights J. 5 2020 84 104 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Interpretationreasoning

    Ethical consciousness can be stated in terms of societal and culture shifts in the awareness of citizens, technology developers and deployers, policy makers, and leaders of industry, in the ethical dimensions of new digital technologies.

    “Drawing particularly on the latter, ethical consciousness can be stated in terms of societal and culture shifts in the awareness of citizens, technology developers and deployers, policy makers, and leaders of industry, in the ethical dimensions of new digital technologies.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Interpretationreasoning

    Digital ethics is a highly interdisciplinary field, requiring expertise that spans the sciences and the humanities.

    “Thus, it is also clear that digital ethics is a highly interdisciplinary field, requiring expertise that spans the sciences (computer science) and the humanities (philosophy, law, sociology, psychology, etc.).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationcitation

    In the context of AI systems there are no robust legal and professional accountability mechanisms, unlike in medicine where a doctor can be struck off or a license withdrawn.

    “and finally, no robust legal and professional accountability mechanisms is present, i.e., in medicine a doctor can be “struck off” and/or a license withdrawn.67”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [67] Mittelstadt B. Principles alone cannot guarantee ethical AI Nat. Machine Intelligence 2019 1 7

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Premisenone given

    Law and ethics are highly interdependent.

    “We do this because we maintain the philosophical position that law and ethics are highly interdependent and because we note that key themes of governance, accountability, and transparency (see section 'Major Themes in AI Ethics' 5) draw heavily on jurisprudence.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given

    The philosophy of technology emerged circa the 1920s and can be read as continuing until today.

    “The philosophy of technology emerged circa the 1920s and can be read as continuing until today (major figures are Martin Heidegger [d. 1976], Herbert Marcuse [d. 1979], and Jurgen Habermas).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The notion that there is a social constructionism/contingency to technology is central to science and technology studies.

    “The notion that there is a social constructionism/contingency to technology is central to science and technology studies.44,45”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [44] Winner L. Do artifacts have politics? Daedalus 109 1980 121 136

      Not checked in the free run.

    • cites [45] Zittrain J.L. The generative internet Harvard Law Journal 119 2006 10.1145/1435417.1435426

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Auditing and impact assessments involve the creation of metrics for tracing and tracking decisions, making the technologies accessible for verification and accountability.

    “Auditing and impact assessments involve the creation of metrics for tracing and tracking decisions, making the technologies accessible for verification and accountability.68”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [68] Koshiyama A. Kazim E. Treleaven P. Rai P. Szpruch L. Pavey G. Towards Algorithm Auditing: A Survey on Managing Legal, Ethical and Technological Risks of AI, ML and Associated Algorithms 2021 SSRN PMC11285902 39076787 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

    Interpretationcitation

    An increase in digital literacy will facilitate the shift in ethical consciousness, and is particularly important for meaningful human intervention and issues of consent.

    “Such a shift will be facilitated through an increase in digital literacy, particularly important for meaningful human intervention and issues of consent.73”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [73] Miller C. Coldicutt R. People, power and technology: the tech workers’ view. Doteveryone https://doteveryone.org.uk/report/workersview 2019

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    In the context of AI systems there is no professional history and norms, because the technologies are still at an early stage and the field of AI ethics is still contested.

    “second, there is no professional history and norms, i.e., the technologies are still at an early stage and the field of AI ethics is still contested (as discussed above);”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisenone given

    AI ethics is a nascent field and a subset of the wider field of digital ethics.

    “It can be read as a nascent field and as a subset of the wider field of digital ethics, which addresses concerns raised by the development and deployment of new digital technologies, such as AI, big data analytics, and blockchain technologies.”
    • Also stated in: § Introduction · plain
      “We read AI ethics as a nascent field and as a subset of the wider field of digital ethics, which addresses concerns raised by the development and deployment of new digital technologies, such as AI, big data analytics, and blockchain technologies.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisereasoning

    The predominant discourse of AI ethics is found within Western European and North American contexts.

    “the predominant discourse of AI ethics is found within the Western European and North American contexts, which is also to be read in terms of our access to the English language literature.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The six themes of human agency and oversight, safety, privacy, transparency, fairness, and accountability can be read as falling under the umbrella of "Trustworthy AI".

    “Drawing on the European Commission's “Ethics Guidelines for Trustworthy AI” (2019),30 we note that these themes can be read as falling under the umbrella of “Trustworthy AI” (introduced above in Concepts and terms).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [30] Ethics Guidelines for Trustworthy AI: High-level Expert Group on Artificial Intelligence (8th April 2019) European Commission.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    AI ethics is inherently interdisciplinary and not exclusively in the purview of philosophers, lawyers, sociologists, or engineers.

    “This also challenges any attempt to silo questions and responsibilities; AI ethics is not exclusively in the purview of philosophers or lawyers or sociologists or engineers, etc., rather it is inherently interdisciplinary.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

    Premisereasoning

    Technologies can be used for good and for bad.

    “In other words, technologies can be used for good and for bad.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

    Interpretationreasoning

    An increase in literacy is crucial in informing all relevant stakeholders and society at large and will be facilitated through a holistic education and training agenda touching upon and integrating ethics, policy, and engineering.

    “as such an increase in “literacy” is crucial in informing all relevant stakeholders and society at large and will be facilitated through a holistic education and training agenda (touching upon and integrating ethics, policy, and engineering).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Premisenone given

    AI engineers may not be best placed to understand and discern the ethical dimension and potential impact of the technology.

    “The idea here is that AI engineers may not be best placed to understand and discern the ethical dimension and potential impact of the technology and, as such, experts from anthropology, sociology, philosophy, psychology, law, etc., i.e., “ethicists,” can be integrated into the team at the development stage.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    There is a distinction between fairness in terms of treatment and fairness in terms of impact.

    “This also includes a distinction between fairness in terms of treatment and fairness in terms of impact.78”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [78] Lipton Z. McAuley J. Chouldechova A. Does mitigating ML’s impact disparity require treatment disparity? Advances in Neural Information Processing Systems 2018 NeurIPS 8125 8135

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

    Premisereasoning

    Ethical AI is a branch of applied ethics and is inherently concerned with how AI systems impact human beings.

    “Ethical AI is a branch of applied ethics and, as such, is inherently concerned with how AI systems impact human beings.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

    Interpretationnone given

    The relationship between AI and the economy will become a major theme of AI ethics.

    “we believe that the relationship between AI and the economy will become a major theme of AI ethics.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Privacy relies on a distinction between the private-personal and the public-political/communal sphere, where the former is seen as demanding a higher level of respect for privacy than the latter.

    “This relies on a distinction between the private-personal and the public-political/communal sphere, where the former is seen as demanding a higher level of respect for privacy than the latter.76”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [76] Mittelstadt B. From individual to group privacy in big data analytics Philos. Technol. 30 2017 475 494

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

    Interpretationnone given

    Communication in an accessible and explicable vernacular to users will facilitate meaningful engagement of wider society with AI systems.

    “•Participation: communication, in an accessible and explicable vernacular, to users will facilitate meaningful engagement of wider society with AI systems.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

    Interpretationnone given

    Accountability is central for the possibility of redress and assigning legal liability.

    “Finally, accountability is central for the possibility of redress and assigning legal liability.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    The relationship between data ethics and AI will increase in importance.

    “given the substantial literature, practice, and regulation around data ethics, we anticipate that the relationship between data ethics and AI will increase in importance.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Interpretationnone given

    Debates about how AI impacts the structure of the state, the notion of a nation, and trust in government communication, management, and service provision will become central themes within AI ethics.

    “debates about how AI impacts the structure of the state, the very notion of a nation (with clear juridical remit), and trust in government communication, management, and service provision, will become central themes within AI ethics.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationreasoning

    Respect for human agency would entail meaningful and informed consent, including the right to withdraw consent and be presented with consent mechanisms that are explicable in the context of an average user.

    “Another issue is consent, where respect for human agency would entail meaningful and informed consent, including the right to withdraw consent and be presented with consent mechanisms that are explicable in the context of an average user.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    A system may have been developed for one use and then be appropriated and/or modified for another, malicious, use (dual use).

    “•Malicious use: a system may have been developed for one use and then be appropriated and/or modified for another, malicious, use, i.e., dual use (e.g., the weaponization of delivery drones).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationreasoning

    A system that is unreliable and does not reproduce results will lead to untrustworthiness in the system.

    “In the context of robustness this is important because a system that is unreliable and does not reproduce results will lead to untrustworthiness in the system.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Interpretationnone given

    Concerns regarding the natural environment and climate will feature strongly in AI ethics.

    “we believe that concerns regarding the natural environment and climate will also feature strongly in AI ethics.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Interpretationnone given

    AI ethics will develop such that it will become clear that the field is inherently interdisciplinary.

    “We believe that AI ethics will develop such that it will become clear that the field is inherently interdisciplinary.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    There are mutually exclusive theories of fairness, such as corrective, distributive, procedural, substantive, and comparative.

    “A central question is what definition(s) of fairness/justice to commit to, i.e., there are mutually exclusive theories of fairness, such as corrective, distributive, procedural, substantive, comparative, etc.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

    Premisenone given

    There are significant benefits that people and society will gain from AI systems.

    “although much of the discourse in AI ethics concerns mitigation of harms, it is also clear that there are significant benefits that people and society will gain from these systems.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

    Interpretationcitation

    The legal status of algorithms with respect to responsibilities and obligations of those developing and deploying them is likely to raise complex questions regarding legal culpability, agency, and personhood, and will increase in complexity and importance as AI systems become more embedded in daily life and more functionally significant.

    “the legal status of an algorithm with respect to responsibilities and obligations of those developing and deploying them, is likely to raise a number of complex questions regarding the nature of legal culpability and even questions of agency and personhood. We anticipate that this question will increase in complexity and importance the more AI systems are embedded in people's daily lives and in”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [84] Zekos G.I. AI and legal issues Economics and Law of Artificial Intelligence 2021 Springer 401 430

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationnone given

    There is not a one-size-fits-all approach to design for accessibility; designs should be user-friendly toward different demographics, cultural and linguistic groups, and those with disabilities.

    “aside from affordability, designs should be user-friendly (e.g., toward different demographics, cultural and linguistic groups, and, in particular, those with disabilities), i.e., there is not a one-size-fits-all approach.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationnone given

    Explicability depends on the technical knowledge of the person, what role they play in the development and deployment of the system, and what kind of end-user they are.

    “i.e., explicability will depend on the technical knowledge of the person, what role they play in the development and deployment of the system, and what kind of end-user they are.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

    Interpretationnone given

    Ensuring human responsibility entails mapping of duties and risks to responsibilities and roles within an institution.

    “Ensuring human responsibility also entails mapping of duties and risks to responsibilities and roles within an institution.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Algorithmic impact assessments can be divided into two approaches: impact assessments and auditing of technology.

    “We can divide this into two approaches, namely (1) impact assessments and (2) auditing of technology.821”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [82] Examining the Black Box: Tools for assessing algorithmic systems. (2020). Ada-Lovelace Institute & DataKind UK. https://www.adalovelaceinstitute.org/examining-the-black-box-tools-for-assessing-algorithmic-systems/.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

    Interpretationnone given

    Openness is crucial to establishing trust and accountability.

    “This theme is based on the principle of openness, which is crucial to establishing trust and accountability.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

    Premisenone given

    AI ethics is a nascent field that is open and plural in its various standpoints.

    “we recognize AI ethics as a nascent field that is open and plural in its various standpoints.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Premisecitation

    There are considerable benefits of AI for people and society.

    “Indeed, although the discussion is presented in terms of mitigating risks and potential harms, it is important to bear in mind the considerable benefits of AI for people and society.75”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [75] Taddeo M. Floridi L. How AI can be a force for good Science 361 2018 751 752 30139858

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Interpretationnone given

    It is not possible to fully anticipate risk.

    “With respect to the latter, it is not possible to fully anticipate risk and, as such, mechanisms can be put into place to mitigate this (e.g., fallback mechanisms, automatic stops—statistical or rule-based, metrics, periodic request for human operators to continue operating, etc.).”
    Quote in the paperNot a substantial claim
67 of 67

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • AI explainability underscores that insights into how AI systems functioned and made a decision should be well informed and explicable to stakeholders, though explicability relies on their technical expertise and role.

    “AI explainability underscores the insights in how AI systems functioned and made a decision should be well informed and explicable to the stakeholders, though the explicability relies on their technical expertise and role (Kazim & Koshiyama, 2021; UNESCO, 2019).”

    From Ethical principles for artificial intelligence in education (2023) · cites this paper as Kazim & Koshiyama, 2021

  • The ethical concerns echo wider AI ethics discourse, particularly in healthcare, and touch on fairness, bias, non-maleficence, transparency, and privacy.

    “These concerns echo the wider discourse on AI ethics86–88, particularly in healthcare89, and touch on issues of fairness, bias, non-maleficence, transparency, and privacy.”

    From The ethics of ChatGPT in medicine and healthcare: a systematic review on large language models (LLMs) (2024) · cites this paper as [86]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • House of Lords report is dated 2018 and 2019

    Minor

    The text dates "Regulating in a Digital World" to 2018, but its reference entry dates it to 2019.

1 of 1

Cites

Flagged first, then by citations

3 of the 76 references couldn’t be matched to a known paper.

Research works printed but not linked: 4

  1. 1993, Vol. 11
  2. Treleaven, 2015, Vol. 3
  3. Zittrain, 2006, The generative internet, Harvard Law Journal, 119
  4. Wang, 2019, 64, 243

Cited by

Sorted by citations

Related

Authors
Kazim, E., & Koshiyama, A. S.
Affiliations
University College London (GB)
Corresponding author
  • Kazim, Emre
  • Koshiyama, Adriano Soares
ORCID
Venue
Patterns
Publisher
Elsevier BV
Published
2021-09
Volume
2
Issue
9
Pages
100314
ISSN
2666-3899 (ISSN-L)
DOI
10.1016/j.patter.2021.100314
PMC
PMC8441585
OpenAlex
W3199717448
Semantic Scholar
8162b7e787b05854904a788eb17045916739cb21
Cited by
269
References
76
Citations by year
2026: 56 · 2025: 74 · 2024: 70 · 2023: 36 · 2022: 21 · 2021: 11 · 2020: 1
FWCI
3.35
Type
journal article
Language
en
Open access
Licence
Locations
  • Patterns (publishedVersion, cc-by-nc-nd)
  • SSRN Electronic Journal (publishedVersion)
  • Europe PMC (PubMed Central) (publishedVersion, other-oa)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • UCL Discovery (University College London) (submittedVersion)
Funders
University College London
Topics
  • Social Sciences › Safety Research › Ethics and Social Impacts of AI
  • Social Sciences › Decision Sciences › Information Systems and Management › Ethics in Business and Education
  • Social Sciences › Sociology and Political Science › Flooding and Environmental Impact
Keywords
AI ethics, digital ethics, engineering ethics, philosophy of technology, science and technology studies, ethical principles, moral consciousness, interdisciplinary research

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review