Putting AI ethics to work: are the tools fit for purpose?

Bias, unfairness and lack of transparency and accountability in Artificial Intelligence (AI) systems, and the potential for the misuse of predictive models for decision-making have raised concerns about the ethical impact and unintended consequences of new technologies for society across every sector where data-driven innovation is taking place. This paper reviews the landscape of suggested ethical frameworks with a focus on those which go beyond high-level statements of principles and offer practical tools for application of these principles in the production and deployment of systems. This work provides an assessment of these practical frameworks with the lens of known best practices for impact assessment and audit of technology. We review other historical uses of risk assessments and audits and create a typology that allows us to compare current AI ethics tools to Best Practices found in previous methodologies from technology, environment, privacy, finance and engineering. We analyse current AI ethics tools and their support for diverse stakeholders and components of the AI development and deployment lifecycle as well as the types of tools used to facilitate use. From this, we identify gaps in current AI ethics tools in auditing and risk assessment that should be considered going forward.

[1]  Hanna M. Wallach,et al.  Co-Designing Checklists to Understand Organizational Challenges and Opportunities around Fairness in AI , 2020, CHI.

[2]  Rachel K. E. Bellamy,et al.  AI Fairness 360: An Extensible Toolkit for Detecting, Understanding, and Mitigating Unwanted Algorithmic Bias , 2018, ArXiv.

[3]  Michael Veale,et al.  Fairness and Accountability Design Needs for Algorithmic Support in High-Stakes Public Sector Decision-Making , 2018, CHI.

[4]  Aaron Marcus,et al.  Seven HCI Grand Challenges , 2019, Int. J. Hum. Comput. Interact..

[5]  R. Freeman.  Strategic Management by R. Edward Freeman , 2010 .

[6]  Nicholas Diakopoulos,et al.  Accountability in algorithmic decision making , 2016, Commun. ACM.

[7]  Kevin B Smith,et al.  Typologies, Taxonomies, and the Benefits of Policy Classification , 2002 .

[8]  Daniel J. Solove.  A Taxonomy of Privacy , 2006 .

[9]  Ortwin Renn.  Risk Governance: Coping with Uncertainty in a Complex World , 2008 .

[10]  Inioluwa Deborah Raji,et al.  Model Cards for Model Reporting , 2018, FAT.

[11]  Laura Forlano,et al.  Participation Is not a Design Fix for Machine Learning , 2020, EAAMO.

[12]  F. Vanclay,et al.  Human rights and impact assessment: clarifying the connections in practice , 2013 .

[13]  Dennis D. Hirsch,et al.  Business Data Ethics: Emerging Trends in the Governance of Advanced Analytics and AI , 2020, SSRN Electronic Journal.

[14]  Daniel J. Solove,et al.  Privacy and Power: Computer Databases and Metaphors for Information Privacy , 2001 .

[15]  Jason Radford,et al.  Theory In, Theory Out: The Uses of Social Theory in Machine Learning for Social Science , 2020, Frontiers in Big Data.

[16]  Charles D. Raab,et al.  Information privacy, impact assessment, and the place of ethics⁎ , 2020, Comput. Law Secur. Rev..

[17]  Kate Crawford,et al.  Atlas of AI , 2021, Perspectives on Science and Christian Faith.

[18]  Luciano Floridi,et al.  Ethics-Based Auditing to Develop Trustworthy AI , 2021, Minds Mach..

[19]  Elettra Bietti,et al.  From ethics washing to ethics bashing: a view on tech ethics from within moral philosophy , 2019, FAT*.

[20]  Daniel J. Solove,et al.  Privacy Harms , 2021, SSRN Electronic Journal.

[21]  Ben Hutchinson,et al.  50 Years of Test (Un)fairness: Lessons for Machine Learning , 2018, FAT.

[22]  Joanna J. Bryson,et al.  The Artificial Intelligence of the Ethics of Artificial Intelligence , 2020 .

[23]  Bernd Carsten Stahl,et al.  Artificial intelligence ethics guidelines for developers and users: clarifying their content and normative implications , 2020, J. Inf. Commun. Ethics Soc..

[24]  Niels Bantilan,et al.  Themis-ml: A Fairness-Aware Machine Learning Interface for End-To-End Discrimination Discovery and Mitigation , 2017, ArXiv.

[25]  D. Pearce.  Cost-benefit analysis , 1971 .

[26]  A. Westin.  Information technology in a democracy , 1971 .

[27]  Peter P.C.C. Verbeek,et al.  Beyond checklists: toward an ethical-constructive technology assessment , 2015 .

[28]  Rickmer Braren,et al.  Secure, privacy-preserving and federated machine learning in medical imaging , 2020, Nature Machine Intelligence.

[29]  C. Starr.  Social benefit versus technological risk. , 1969, Science.

[30]  Madhulika Srikumar,et al.  Principled Artificial Intelligence: Mapping Consensus in Ethical and Rights-Based Approaches to Principles for AI , 2020, SSRN Electronic Journal.

[31]  L. Preston,et al.  The Stakeholder Theory of the Corporation: Concepts, Evidence, and Implications , 1995 .

[32]  C. Robb.  Poverty and Social Impact Analysis - Linking Macroeconomic Policies to Poverty Outcomes: Summary of Early Experiences , 2003, SSRN Electronic Journal.

[33]  Eevi E. Beck,et al.  P for Political: Participation is Not Enough , 2002 .

[34]  A. Hoffmann.  Where fairness fails: data, algorithms, and the limits of antidiscrimination discourse , 2019, Information, Communication & Society.

[35]  Sven Ove Hansson,et al.  The case for ethical technology assessment (eTA) , 2006 .

[36]  B. Roessler,et al.  New Ways of Thinking about Privacy , 2006 .

[37]  H Roberts,et al.  Risk Society: Towards a New Modernity , 1994 .

[38]  Dallas Card,et al.  The Values Encoded in Machine Learning Research , 2021, ArXiv.

[39]  Joseph Gray Jackson,et al.  Privacy and Freedom , 1968 .

[40]  Virginia E. Eubanks.  Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor , 2018 .

[41]  M. Patton.  Qualitative Research & Evaluation Methods: Integrating Theory and Practice , 2014 .

[42]  P. Driessen,et al.  Public participation in environmental impact assessment: why, who and how? , 2013 .

[43]  P. Missier,et al.  Capturing and querying fine-grained provenance of preprocessing pipelines in data science , 2020, Proc. VLDB Endow..

[44]  Mariarosaria Taddeo,et al.  The ethics of algorithms: Mapping the debate , 2016, Big Data Soc..

[45]  Nicholas Diakopoulos,et al.  Accountability in Algorithmic Decision-making , 2015 .

[46]  Jess Whittlestone,et al.  The Role and Limits of Principles in AI Ethics: Towards a Focus on Tensions , 2019, AIES.

[47]  Jesper Simonsen,et al.  Routledge International Handbook of Participatory Design , 2012 .

[48]  Luciano Floridi,et al.  Formalising trade-offs beyond algorithmic fairness: lessons from ethical philosophy and welfare economics , 2021, AI Ethics.

[49]  Thomas Clarke,et al.  Accounting for Enron: Shareholder Value and Stakeholder Interests , 2005 .

[50]  Francesca Rossi,et al.  AI4People—An Ethical Framework for a Good AI Society: Opportunities, Risks, Principles, and Recommendations , 2018, Minds and Machines.

[51]  Joseph F. Coates,et al.  Some methods and techniques for comprehensive impact assessment , 1974 .

[52]  Heidy Khlaaf,et al.  Disruptive Innovations and Disruptive Assurance: Assuring Machine Learning and Autonomy , 2019, Computer.

[53]  Anna Jobin,et al.  The global landscape of AI ethics guidelines , 2019, Nature Machine Intelligence.

[54]  Ville Vakkuri,et al.  Ethically Aligned Design of Autonomous Systems: Industry viewpoint and an empirical study , 2019, ArXiv.

[55]  Hanna M. Wallach,et al.  Fairlearn: A toolkit for assessing and improving fairness in AI , 2020 .

[56]  Michelle Seng Ah Lee,et al.  The Landscape and Gaps in Open Source Fairness Toolkits , 2021, CHI.

[57]  Clinton Free,et al.  Hybridized Professional Groups and Institutional Work: COSO and the Rise of Enterprise Risk Management , 2014 .

[58]  J. Borges,et al.  A TAXONOMY OF PRIVACY , 2006 .

[59]  Peter Henderson,et al.  Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims , 2020, ArXiv.

[60]  D W Pennington,et al.  Life cycle assessment: Part 1: Framework, goal and scope definition, inventory analysis, and applications , 2004 .

[61]  Leonhard Hennen,et al.  Why do we still need participatory technology assessment? , 2012, Poiesis Prax..

[62]  Marcello Ienca,et al.  Artificial Intelligence: the global landscape of ethics guidelines , 2019, ArXiv.

[63]  James McConvill,et al.  Principles of Contemporary Corporate Governance , 2005 .

[64]  John Rushby,et al.  The Interpretation and Evaluation of Assurance Cases , 2015 .

[65]  Ortwin Renn,et al.  Global Risk Governance , 2008 .

[66]  J. Borenstein,et al.  AI Ethics in the Public, Private, and NGO Sectors: A Review of a Global Document Collection , 2021, IEEE Transactions on Technology and Society.

[67]  Mariette Bengtsson,et al.  How to plan and perform a qualitative study using content analysis , 2016 .

[68]  Terje Aven,et al.  Risk assessment and risk management: Review of recent advances on their foundation , 2016, Eur. J. Oper. Res..

[69]  M. Edwards,et al.  Prospects and Perils of Fiscal Impact Analysis , 2009 .

[70]  Roger Clarke,et al.  Privacy impact assessment: Its origins and development , 2009, Comput. Law Secur. Rev..

[71]  Robert V. O'Neill,et al.  Treatment of risk in environmental impact assessment , 1987 .

[72]  Thilo Hagendorff,et al.  The Ethics of AI Ethics: An Evaluation of Guidelines , 2019, Minds and Machines.

[73]  R. Schwing,et al.  Interdisciplinary Vision: The First 25 Years of the Society for Risk Analysis (SRA), 1980–2005 , 2005, Risk analysis : an official publication of the Society for Risk Analysis.

[74]  Richard K. Morgan.  Environmental impact assessment: the state of the art , 2012 .

[75]  Kristina Lerman,et al.  A Survey on Bias and Fairness in Machine Learning , 2019, ACM Comput. Surv..

[76]  Madeleine Clare Elish,et al.  Algorithmic Impact Assessments and Accountability: The Co-construction of Impacts , 2021, FAccT.

[77]  R. Freeman.  Strategic Management: A Stakeholder Approach , 2010 .

[78]  Luciano Floridi,et al.  From What to How: An Initial Review of Publicly Available AI Ethics Tools, Methods and Research to Translate Principles into Practices , 2019, Science and Engineering Ethics.

[79]  Luke Stark,et al.  Better, Nicer, Clearer, Fairer: A Critical Assessment of the Movement for Ethical Artificial Intelligence and Machine Learning , 2019, HICSS.

[80]  Klaus Krippendorff,et al.  Content Analysis: An Introduction to Its Methodology , 1980 .

[81]  P. Mayring.  Qualitative Content Analysis: Demarcation, Varieties, Developments , 2019 .