Back

Principles alone cannot guarantee ethical AI

Brent Mittelstadt

article2019Nature Machine Intelligence1,540 citations

Explains why high-level ethical principles modeled on medical ethics fail to govern artificial intelligence effectively due to critical structural differences in accountability, professional norms, and practical implementation.

Listen

The rapid expansion of artificial intelligence across critical public and private sectors has generated intense international debate regarding its ethical governance. At least 84 public-private initiatives have published high-level ethical frameworks, tenets, and principles to guide technology design and deployment. Many of these frameworks converge around four core concepts that closely resemble traditional biomedical ethics: autonomy, harm prevention, fairness, and explicability. The article evaluates whether this principle-based approach can successfully govern artificial intelligence, comparing the field directly with the historical and structural mechanisms of medical ethics.

To conduct this evaluation, the article reviews the outputs of major artificial intelligence ethics initiatives alongside established literature on biomedical principlism, professional governance, and software engineering. It identifies four fundamental structural differences that impede artificial intelligence ethics: a lack of common aims and fiduciary duties, an absence of professional history and shared behavioral norms, a deficit of proven methodologies to translate abstract principles into daily development workflows, and a shortage of binding legal and professional accountability mechanisms.

The findings indicate that current ethical statements remain too abstract to guide real-world decision-making. Concepts such as fairness and human dignity are inherently contested and subject to conflicting interpretations, which allows high-level consensus to obscure serious political and normative disagreements. Furthermore, private-sector artificial intelligence developers face commercial incentives that prioritize shareholder value and cost reduction over public welfare. Unlike medicine, artificial intelligence development lacks licensing requirements, established standards of care, formal disciplinary bodies, and malpractice frameworks. Consequently, self-regulatory voluntary codes have shown no measurable impact on practitioner conduct unless paired with active enforcement and embedded cultural accountability.

These findings suggest that relying solely on abstract principles creates a misleading impression of safety while potentially delaying necessary statutory regulation. In commercial environments, ethical initiatives risk functioning as superficial virtue signaling rather than genuine governance. Complex social and political challenges cannot be reduced to simple technological fixes or engineering parameters, particularly when technical decisions are made in private by multi-disciplinary, distributed teams where the long-term impact on affected individuals is opaque.

The article recommends five key actions for policy-makers, industry leaders, and researchers: establish clear pathways to impact with transparent ethical auditing and dataset documentation; support bottom-up, case-based ethics research within commercial development environments; license practitioners who develop high-risk applications, such as public-sector surveillance systems; shift governance focus from individual developer behavior to broader organizational business models; and treat ethics as an ongoing, deliberative process rather than a static technical solution.

The assessment is limited by the fast-evolving landscape of global technology policy, the diverse technical architectures within the field, and existing regulatory variations across jurisdictions. Nonetheless, there is strong confidence that voluntary high-level principles alone are insufficient. Stakeholders must combine principled guidance with structural accountability, legal oversight, and rigorous organizational enforcement to achieve reliable, trustworthy artificial intelligence systems.

The Alan Turing InstituteUniversity of Oxfordartificial intelligenceethical AIethical principlescomputer ethics
arXiv: 1906.06668
  • Paper: The global landscape of AI ethics guidelines, Anna Jobin et al. (2019). This scoping review of 84 AI ethics guidelines establishes the empirical landscape and convergence of principles that the source paper directly examines and critiques.
  • Paper: Fairness and Abstraction in Sociotechnical Systems, Andrew D. Selbst et al. (2019). This paper analyzes why abstract technical formulations fail in complex sociotechnical contexts, providing the foundational conceptual framing for why high-level AI principles struggle in practice.
Cover for Principles alone cannot guarantee ethical AI

Abstract

AI Ethics is now a global topic of discussion in academic and policy circles. At least 84 public-private initiatives have produced statements describing high-level principles, values, and other tenets to guide the ethical development, deployment, and governance of AI. According to recent meta-analyses, AI Ethics has seemingly converged on a set of principles that closely resemble the four classic principles of medical ethics. Despite the initial credibility granted to a principled approach to AI Ethics by the connection to principles in medical ethics, there are reasons to be concerned about its future impact on AI development and governance. Significant differences exist between medicine and AI development that suggest a principled approach in the latter may not enjoy success comparable to the former. Compared to medicine, AI development lacks (1) common aims and fiduciary duties, (2) professional history and norms, (3) proven methods to translate principles into practice, and (4) robust legal and professional accountability mechanisms. These differences suggest we should not yet celebrate consensus around high-level principles that hide deep political and normative disagreement.

Topics

artificial intelligence

Keywords

ethical AIethical principlescomputer ethicsprofessional ethicsethical decision makingaccountability mechanismsself-regulatory codesethics guidelinesprofessional standardsOECD Principles

Table of Contents

  • 1 Introduction
  • 2.2 Professional history and norms
  • Acknowledgements
  • Competing Interests
  • References

Knowls

  1. Knowl 1 — Comparative Framework: The Four Structural Deficits of Principlism in AI Development

    theoretical result

    While global artificial intelligence (AI) ethics initiatives have converged on high-level principles that mirror the four classical principles of biomedical ethics (respect for autonomy, prevention of harm, fairness, and explicability), high-level principlism cannot be successfully transplanted into AI development without structural adaptation. In medicine, principlism functions effectively because it is anchored in an established, centuries-old professional ecosystem. AI development lacks this foundation and exhibits four structural deficits:

    1. Absence of common aims and fiduciary duties: AI developers do not share a foundational commitment to prioritize the public interest or user well-being over commercial goals.
    2. Lack of professional history and shared norms: AI development lacks a cohesive professional identity, historical ethical milestones, and a recognized standard of care.
    3. Absence of proven translation methods: There are no established, empirically validated methodologies for translating abstract principles into low-level software requirements and system architectures.
    4. Absence of legal and professional accountability: The field lacks mandatory licensing, enforceable professional disciplinary systems, and legal liability mechanisms equivalent to medical malpractice.

  2. Knowl 2 — Structural Deficit of Common Aims and Fiduciary Duties in AI Development

    theoretical result

    In medicine, practitioners belong to a moral community united by an overarching fiduciary commitment to promote patient health and well-being. This primary duty takes precedence over institutional, managerial, and financial interests, establishing a cooperative environment for ethical decision-making.

    In contrast, artificial intelligence systems are predominantly engineered in private, commercial settings where companies and employees owe primary fiduciary duties to corporate shareholders. AI development does not serve an analogue to a "patient" whose vital interests are granted default primacy. Without legal frameworks establishing fiduciary duties or public service commitments toward data subjects and users, ethical decision-making becomes a competitive negotiation between private commercial incentives (speed, cost-cutting, profitability) and public welfare.

  3. Knowl 3 — Structural Deficit of Professional History, Shared Culture, and Definitional Precision in AI

    theoretical result

    Unlike medicine, which has developed an evolving professional standard of care through historical milestones (such as the Hippocratic tradition, the Declaration of Helsinki, the Nuremberg Code, and the Belmont Report), AI development lacks a unified history, culture, or professional standard. Three major factors impede the establishment of such norms in AI:

    • Spatial and Temporal Distance: Clinical decisions typically yield observable effects on physical bodies, whereas algorithmic harms are often continuous, indirect, delayed, and geographically separated from developers, which empirically increases the likelihood of unethical conduct.
    • Opacity and Distributed Responsibility: Modern machine learning models and large multi-disciplinary, multi-national engineering teams disperse agency across complex networks, making it difficult to predict model behaviors or trace ethical failures back to individual choices.
    • Essentially Contested Concepts: Principles in AI ethics rely heavily on abstract concepts such as "fairness," "dignity," and "transparency." These are essentially contested concepts whose operational definitions depend on background political and philosophical commitments. Consensus on high-level principle statements conceals underlying normative disagreement rather than resolving it.

  4. Knowl 4 — Structural Deficit of Translation Methodologies from Principles to System Design

    theoretical result

    High-level ethical principles cannot be deductively derived into mid-level norms or low-level technical specifications. Every phase of specification requires independent normative justification and trade-offs that cannot be settled by high-level consensus alone.

    Existing value-conscious engineering frameworks (such as Value-Sensitive Design, Participatory Design, and Values@Play) face two fundamental limitations when applied to commercial AI:

    1. Procedural vs. Functional Mismatch: These methods introduce stakeholders and values into the development process, but they do not provide technical mechanisms to inject specific values into system architectures or measure the degree to which an artifact embodies those values.
    2. Commercial Friction: Stakeholder engagement, embedded ethicists, and resolving conflicting interpretations of contested concepts introduce project overhead and delays. In commercial environments governed by speed and efficiency, these procedural requirements are routinely discarded when they conflict with business objectives.

  5. Knowl 5 — Structural Deficit of Legal and Professional Accountability Mechanisms in AI Engineering

    theoretical result

    In traditional professions, ethical codes are reinforced by legal and regulatory frameworks, including state licensing boards, malpractice liability, and mandatory peer review bodies that hold the authority to revoke a practitioner's license to practice.

    In software engineering and AI development, voluntary codes of ethics (such as those of the ACM and IEEE) exert little discernible influence on day-to-day practitioner decision-making without active organizational enforcement and legal backing. Professional bodies possess no formal sanctioning power beyond organizational expulsion, which does not affect a developer's legal ability to practice. In the absence of external punitive mechanisms and legal standards of care, voluntary self-regulatory codes risk serving as superficial virtue signaling that delays binding regulation while offering false assurances of algorithmic trustworthiness.

  6. Knowl 6 — Constitutive Characteristics of a Formal Profession in Applied Ethics

    data/table

    A formal profession is defined by five structural criteria that enable self-governance and the effective enforcement of ethical standards across its practitioners:

    Characteristic Core Definition and Scope
    Specialised education and training Members undergo extensive specialized training, typically via accredited degree programmes.
    Commitment to public service Practitioners publicly declare that specialized expertise serves society, taking precedence over private gain.
    Higher standard of care Members commit to upholding higher ethical standards than ordinary commercial relationships.
    Enforcement and self-governance Disciplinary systems administered by professional associations actively enforce codes and penalize misconduct.
    Licensing Entry is legally restricted via government-sanctioned licensure to protect the public from harm.

    AI and software development currently meet none of these five conditions: entry is unlicensed, public service holds no legal precedence over corporate fiduciary duties, no legally defined standard of care exists, and associations lack punitive disciplinary powers over practitioners' livelihoods.

  7. Knowl 7 — Diagnostic Framework for Evaluating AI Ethics Charters and Principle Statements

    model/method

    To evaluate whether an artificial intelligence ethics guideline or charter provides actionable governance rather than superficial rhetoric, the following eight diagnostic criteria must be assessed:

    Dimension Diagnostic Question
    1. Authorship Who wrote the principles, and through what process?
    2. Purpose Who is the statement intended for, and what is its explicit objective?
    3. Obligation Why should a practitioner or organization follow it?
    4. Operationalization How are principles translated and implemented in practical workflows?
    5. Conflict Resolution How should conflicting interpretations of essentially contested concepts be resolved?
    6. Verification How will adherence and compliance be independently monitored?
    7. Enforcement What formal sanctions or consequences occur upon non-compliance?
    8. Recourse How can practitioners raise disagreements or seek normative clarification?

  8. Knowl 8 — Five-Point Governance Strategy for Actionable and Accountable AI Ethics

    model/method

    To overcome the limitations of abstract principlism and establish substantive AI governance, five policy and organizational strategies should be pursued:

    1. Define Sustainable Pathways to Impact: Establish sectoral accountability mechanisms requiring inclusive design, transparent ethical review, standardized model and dataset documentation, and independent third-party ethical auditing.
    2. Support Bottom-Up AI Ethics in the Private Sector: Complement top-down principles with empirical case studies of production AI systems, creating open repositories of algorithmic harms and giving multi-disciplinary researchers access to commercial development environments.
    3. License Developers of High-Risk AI: Institute mandatory professional licensing for engineers developing systems deployed in high-risk or public-service settings (e.g., facial recognition in policing), reattaching professional responsibility to automated decision systems.
    4. Shift from Professional to Organisational Ethics: Expand governance focus from individual developer conduct to institutional business models and corporate structures, ensuring organizational accountability for systemic harms.
    5. Pursue Ethics as an Ongoing Process, Not Technological Solutionism: Reject the assumption that normative and political conflicts can be solved via narrow technical fixes (e.g., computable fairness metrics). AI ethics must be treated as an ongoing, reflexive, and democratic process rather than a static checklist.

Coverage note — No substantial contributed material was omitted; all core critical arguments, diagnostic tables, comparative professional frameworks, and governance recommendations are captured.

References

  1. 1.Greene, D., Hoffmann, A. L. & Stark, L. Better, Nicer, Clearer, Fairer: A Critical Assessment of the Movement for Ethical Artificial Intelligence and Machine Learning. 10 (2019).
  2. 2.Whittaker, M. et al. AI Now Report 2018. (AI Now Institute, 2018).
  3. 3.Jobin, A., Ienca, M. & Vayena, E. The global landscape of AI ethics guidelines. Nat. Mach. Intell. 1, 389–399 (2019).
  4. 4.Nemitz, P. Constitutional democracy and technology in the age of artificial intelligence. Philos. Trans. R. Soc. Math. Phys. Eng. Sci. 376, 20180089 (2018).
  5. 5.Hagendorff, T. The Ethics of AI Ethics -- An Evaluation of Guidelines. ArXiv190303425 Cs Stat (2019).
  6. 6.Calo, R. Artificial Intelligence Policy: A Primer and Roadmap Symposium - Future-Proofing Law: From RDNA to Robots (Part 2). UC Davis Law Rev. 51, 399–436 (2017).
  7. 7.Whittlestone, J., Nyrup, R., Alexandrova, A., Dihal, K. & Cave, S. Ethical and societal implications of algorithms, data, and artificial intelligence: a roadmap for research. 59 (Nuffield Foundation, 2019).
  8. 8.Floridi, L. et al. AI4People—An Ethical Framework for a Good AI Society: Opportunities, Risks, Principles, and Recommendations. Minds Mach. 28, 689–707 (2018).
  9. 9.OECD. Forty-two countries adopt new OECD Principles on Artificial Intelligence - OECD. Forty-two countries adopt new OECD Principles on Artificial Intelligence (2019). Available at: http://www.oecd.org/science/forty-two-countries-adopt-new-oecd-principles-on-artificial-intelligence.htm. (Accessed: 22nd May 2019)
  10. 10.High Level Expert Group on Artificial Intelligence. Ethics Guidelines for Trustworthy AI. (European Commission, 2019).
  11. 11.Filipović, A., Koska, C. & Paganini, C. Developing a Professional Ethics for Algorithmists: Learning from the Examples of Established Ethics. (Bertelsmann Stiftung, 2018).
  12. 12.Beauchamp, T. L. & Childress, J. F. Principles of biomedical ethics. (Oxford University Press, 2009).
  13. 13.Bosk, C. L. Bioethics, Raw and Cooked: Extraordinary Conflict and Everyday Practice. J. Health Soc. Behav. 51, S133–S146 (2010).
  14. 14.Beauchamp, T. L. & DeGrazia, D. Principles and Principlism. in Handbook of Bioethics:: Taking Stock of the Field from a Philosophical Perspective (ed. Khushf, G.) 78, 55–74 (Springer Science & Business Media, 2006).
  15. 15.Marshall, T. H. The recent history of professionalism in relation to social structure and social policy. Can. J. Econ. Polit. Sci. Can. Econ. Sci. Polit. 5, 325–340 (1939).
  16. 16.Frankel, M. S. Professional codes: Why, how, and with what impact? J. Bus. Ethics 8, 109–115 (1989).
  17. 17.Black’s law dictionary. (West, 2009).
  18. 18.MacIntyre, A. After Virtue: A Study in Moral Theory. (Gerald Duckworth & Co Ltd, 2007).
  19. 19.Gillon, R. Do doctors owe a special duty of beneficence to their patients? J. Med. Ethics 12, 171–173 (1986).
  20. 20.Pellegrino, E. D. & Thomasma, D. C. The virtues in medical practice. (Oxford University Press, 1993).
  21. 21.Van den Bergh, J., Deschoolmeester, D., Deschoolmeester, D. & Vlerick Leuven Gent Management School, Gent, Belgium. Ethical Decision Making in ICT: Discussing the Impact of an Ethical Code of Conduct. Commun. IBIMA 1–11 (2010). doi:10.5171/2010.127497
  22. 22.Manders-Huits, N. & Zimmer, M. Values and Pragmatic Action: The Challenges of Introducing Ethical Intel- ligence in Technical Design Communities. 10, 8 (2009).
  23. 23.McDowell, B. Ethical Conduct and the Professional’s Dilemma: Choosing Between Service and Success. (Quorum Books, 1991).
  24. 24.Iacovino, L. Ethical Principles and Information Professionals: Theory, Practice and Education. Aust. Acad. Res. Libr. 33, 57–74 (2002).
  25. 25.Boddington, P. Towards a code of ethics for artificial intelligence research. (Springer Berlin Heidelberg, 2017).
  26. 26.Zarsky, T. Incompatible: The GDPR in the Age of Big Data. Seton Hall Law Rev. 47, (2017).
  27. 27.Wachter, S. & Mittelstadt, B. D. A right to reasonable inferences: re-thinking data protection law in the age of Big Data and AI. Columbia Bus. Law Rev. 2019, (2019).
  28. 28.Balkin, J. M. Information Fiduciaries and the First Amendment. UCDL Rev 49, 1183 (2015).
  29. 29.Kish-Gephart, J. J., Harrison, D. A. & Treviño, L. K. Bad apples, bad cases, and bad barrels: Meta-analytic evidence about sources of unethical decisions at work. J. Appl. Psychol. 95, 1–31 (2010).
  30. 30.Wakabayashi, D. & Shane, S. Google Will Not Renew Pentagon Contract That Upset Employees. The New York Times (2018).
  31. 31.Conger, K. & Wakabayashi, D. Google Employees Protest Secret Work on Censored Search Engine for China. The New York Times (2018).
  32. 32.Wong, J. C. Demoted and sidelined: Google walkout organizers say company retaliated. The Guardian (2019).
  33. 33.Parker, D. B. Ethical Conflicts in Computer Science and Technology. (AFIPS Press, 1981).
  34. 34.Carrese, J. A. & Sugarman, J. The Inescapable Relevance of Bioethics for the Practicing Clinician. Chest 130, 1864–1872 (2006).
  35. 35.Brotherton, S., Kao, A. & Crigger, B. J. Professing the Values of Medicine: The Modernized AMA Code of Medical Ethics. JAMA 316, 1041–1042 (2016).
  36. 36.Greenfield, B. & Jensen, G. M. Beyond a code of ethics: phenomenological ethics for everyday practice. Physiother. Res. Int. n/a-n/a (2010). doi:10.1002/pri.481
  37. 37.Panensky, S. A. & Jones, R. Does IT Go Without Saying? Prof. Times Summer 2018, 30–33 (2018).
  38. 38.Perlman, D. T. Who Pays the Price of Computer Software Failure Notes and Comments. Rutgers Comput. Technol. Law J. 24, 383–416 (1998).
  39. 39.Mittelstadt, B., Allo, P., Taddeo, M., Wachter, S. & Floridi, L. The ethics of algorithms: Mapping the debate. Big Data Soc. 3, (2016).
  40. 40.Jones, T. M. Ethical Decision Making by Individuals in Organizations: An Issue-Contingent Model. Acad. Manage. Rev. 16, 366–395 (1991).
  41. 41.Metcalf, J. & Crawford, K. Where are human subjects in Big Data research? The emerging ethics divide. Big Data Soc. 3, 2053951716650211 (2016).
  42. 42.Burrell, J. How the Machine ‘Thinks:’ Understanding Opacity in Machine Learning Algorithms. Big Data Soc. (2016). doi:10.1177/2053951715622512
  43. 43.Floridi, L. Faultless responsibility: on the nature and allocation of moral responsibility for distributed moral actions. Philos. Trans. R. Soc. Math. Phys. Eng. Sci. 374, 20160112 (2016).
  44. 44.Awad, E. et al. The Moral Machine experiment. Nature 563, 59 (2018).
  45. 45.Ess, C. Ethical pluralism and global information ethics. Ethics Inf. Technol. 8, 215–226 (2006).
  46. 46.van den Hoven, J. Computer Ethics and Moral Methodology. Metaphilosophy 28, 234–248 (1997).
  47. 47.Gallie, W. B. Essentially Contested Concepts. Proc. Aristot. Soc. 56, 167–198 (1955).
  48. 48.Richardson, H. S. Specifying Norms as a Way to Resolve Concrete Ethical Problems. Philos. Public Aff. 19, 279–310 (1990).
  49. 49.Turner, L. Bioethics in a Multicultural World: Medicine and Morality in Pluralistic Settings. 19 (2003).
  50. 50.Rhodes, R. Good and not so good medical ethics. J. Med. Ethics 41, 71–74 (2015).
  51. 51.Clouser, K. D. & Gert, B. A Critique of Principlism. J. Med. Philos. 15, 219–236 (1990).
  52. 52.Degrazia, D. Moving Forward in Bioethical Theory: Theories, Caes, and Specified Principlism. 17, 29 (1992).
  53. 53.Orentlicher, D. The Influence of a Professional Organization on Physician Behavior Symposium on the Legal and Ethical Implications of Innovative Medical Technology. Albany Law Rev. 57, 583–606 (1993).
  54. 54.Papadakis, M. A. et al. Disciplinary Action by Medical Boards and Prior Behavior in Medical School. N. Engl. J. Med. 353, 2673–2682 (2005).
  55. 55.Toulmin, S. How medicine saved the life of ethics. Perspect. Biol. Med. 25, 736–750 (1982).
  56. 56.van de Poel, I. Translating Values into Design Requirements. in Philosophy and Engineering: Reflections on Practice, Principles and Process (eds. Michelfelder, D. P., McCarthy, N. & Goldberg, D. E.) 15, 253–266 (Springer Netherlands, 2013).
  57. 57.Gillon, R. Defending the four principles approach as a good basis for good medical practice and therefore for good medical ethics. J. Med. Ethics 41, 111–116 (2015).
  58. 58.Friedman, B., Hendry, D. G. & Borning, A. A Survey of Value Sensitive Design Methods. Found. Trends® Human–Computer Interact. 11, 63–125 (2017).
  59. 59.Friedman, B. & Kahn, P. H. Human agency and responsible computing: Implications for computer system design. J. Syst. Softw. 17, 7–14 (1992).
  60. 60.Morley, J., Floridi, L., Kinsey, L. & Elhalal, A. From What to How. An Overview of AI Ethics Tools, Methods and Research to Translate Principles into Practices. ArXiv190506876 Cs (2019).
  61. 61.Shilton, K. Values Levers: Building Ethics into Design. Sci. Technol. Hum. Values 38, 374–397 (2013).
  62. 62.MacKinnon, K. S. Computer Malpractice: Are Computer Manufacturers, Service Burreaus, and Programmers Really the Professionals They Claim to Be. St. Clara Rev 23, 1065 (1983).
  63. 63.Bynum, T. W. Ethical decision making and case analysis in computer ethics. in Computer ethics and professional responsibility (eds. Bynum, T. W. & Rogerson, S.) 60–87 (Blackwell, 2004).
  64. 64.Ladd, J. The quest for a code of professional ethics: an intellectual and moral confusion. in Ethical issues in the use of computers (eds. Johnson, D. G. & Snapper, J. W.) 8–13 (Wadsworth Publ. Co., 1985).
  65. 65.McNamara, A., Smith, J. & Murphy-Hill, E. Does ACM’s code of ethics change ethical decision making in software development? in Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering - ESEC/FSE 2018 729–733 (ACM Press, 2018). doi:10.1145/3236024.3264833
  66. 66.Brief, A. P., Dukerich, J. M., Brown, P. R. & Brett, J. F. What’s wrong with the treadway commission report? Experimental analyses of the effects of personal values and codes of conduct on fraudulent financial reporting. J. Bus. Ethics 15, 183–198 (1996).
  67. 67.Helin, S. & Sandström, J. An Inquiry into the Study of Corporate Codes of Ethics. J. Bus. Ethics 75, 253–271 (2007).
  68. 68.McCabe, D. L., Trevino, L. K. & Butterfield, K. D. The Influence of Collegiate and Corporate Codes of Conduct on Ethics-Related Behavior in the Workplace. Bus. Ethics Q. 6, 461–476 (1996).
  69. 69.Jin, K. G., Drozdenko, R. & Bassett, R. Information Technology Professionals’ Perceived Organizational Values and Managerial Ethics: An Empirical Study. J. Bus. Ethics 71, 149–159 (2007).
  70. 70.Shilton, K. “That’s Not An Architecture Problem!”: Techniques and Challenges for Practicing Anticipatory Technology Ethics. 7 (2015).
  71. 71.Goertzel, K. M. Legal liability for bad software. CrossTalk 23, (2016).
  72. 72.Laplante, P. A. Licensing professional software engineers: seize the opportunity. Commun ACM 57, 38–40 (2014).
  73. 73.Pour, G., Griss, M. L. & Lutz, M. The push to make software engineering respectable. Computer 33, 35–43 (2000).
  74. 74.Seidman, S. B. The Emergence of Software Engineering Professionalism. in E-Government Ict Professionalism and Competences Service Science (eds. Mazzeo, A., Bellini, R. & Motta, G.) 280, 59–67 (Springer US, 2008).
  75. 75.Abbott, A. Professional Ethics. Am. J. Sociol. 32 (1983).
  76. 76.O’Connor, J. E. Computer Professionals: The Need for State Licensing. Jurimetr. J. 18, 256–267 (1978).
  77. 77.Gotterbarn, D. The Ethical Computer Practitioner—Licensing the Moral Community: A Proactive Approach. SIGCSE Bull 30, 8–10 (1998).
  78. 78.Suchman, L. Corporate Accountability. Robot Futures (2018).
  79. 79.Angwin, J., Larson, J. & Kirchner, L. Machine Bias. ProPublica (2016).
  80. 80.European Group on Ethics in Science and New Technologies. Statement on Artificial Intelligence, Robotics and ‘Autonomous’ Systems. (European Commission, 2018).
  81. 81.Holstein, K., Vaughan, J. W., Daumé III, H., Dudík, M. & Wallach, H. Improving fairness in machine learning systems: What do industry practitioners need? ArXiv181205239 Cs (2018). doi:10.1145/3290605.3300830
  82. 82.Benkler, Y. Don’t let industry write the rules for AI. Nature 569, 161–161 (2019).
  83. 83.The IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems. Ethically Aligned Design. (IEEE, 2019). doi:10.1007/978-3-030-12524-0_2

Citation

MLA
Mittelstadt, B. “Principles Alone Cannot Guarantee Ethical AI”. Nature Machine Intelligence, vol. 1, no. 11, 2019, pp. 501–07, https://doi.org/10.1038/s42256-019-0114-4.
APA
Mittelstadt, B. (2019). Principles alone cannot guarantee ethical AI. Nature Machine Intelligence, 1(11), 501–507. https://doi.org/10.1038/s42256-019-0114-4
Chicago
Mittelstadt, B. 2019. “Principles Alone Cannot Guarantee Ethical AI”. Nature Machine Intelligence 1 (11): 501–7. https://doi.org/10.1038/s42256-019-0114-4.
Harvard
Mittelstadt, B. (2019) “Principles alone cannot guarantee ethical AI”, Nature Machine Intelligence, 1(11), pp. 501–507. Available at: https://doi.org/10.1038/s42256-019-0114-4.
Vancouver
1. Mittelstadt B (2019) Principles alone cannot guarantee ethical AI. Nature Machine Intelligence 1:501–507

BibTeX

@article{Mittelstadt_2019, title={Principles alone cannot guarantee ethical AI}, volume={1}, ISSN={2522-5839}, url={http://dx.doi.org/10.1038/s42256-019-0114-4}, DOI={10.1038/s42256-019-0114-4}, number={11}, journal={Nature Machine Intelligence}, publisher={Springer Science and Business Media LLC}, author={Mittelstadt, Brent}, year={2019}, month=Nov, pages={501–507} }
Metadata:Crossref

Access the Paper

This paper is available from its original source. Click below to access the PDF.

Open PDF