Richmond Y. Wong, Michael A. Madaio, and Nick Merrill, “Seeing Like a Toolkit: How Toolkits Envision the Work of AI Ethics,” Proceedings of the ACM on Human-Computer Interaction, vol. 7. doi: 10.1145/3579621
Abstract: Numerous toolkits have been developed to support ethical AI development. However, toolkits, like all tools, encode assumptions in their design about what work should be done and how. In this paper, the authors conduct a qualitative analysis of 27 AI ethics toolkits to critically examine how the work of ethics is imagined and how it is supported by these toolkits. Specifically, the team examines the discourses toolkits rely on when talking about ethical issues, who they imagine should do the work of ethics, and how they envision the work practices involved in addressing ethics. Among the toolkits, they identify a mismatch between the imagined work of ethics and the support the toolkits provide for doing that work. In particular, the authors identify a lack of guidance around how to navigate labor, organizational, and institutional power dynamics as they relate to performing ethical work. They use these omissions to chart future work for researchers and designers of AI ethics toolkits.
The paper focuses on the following research questions:
- What are the discourses of ethics that ethical AI toolkits draw on to legitimize their use?
- Who do the toolkits imagine as doing the work of addressing ethics in AI?
- What do toolkits imagine to be the specific work practices of addressing ethics in AI?
To find out more about the answers, read the full paper here: doi.org/10.1145/3579621
The following toolkits are included in the review:
- Ethics Kit (Open Data Institute, Common Good, Coop Digital, Hyper Island, Plot )
Design Exercises, Worksheets — for Designers - Model Cards (Google)
Examples, Webpage — for Developers, Policymakers, Analysts, Advocates, Users
- AI Fairness 360 (IBM)
Open Source Code, Documentation, Code Examples, Tutorials — for Data Scientists
- InterpretML (Microsoft)
Open Source Code, Documentation, Code Examples — for Data Scientists
- Fairlearn (Miro Dudik (Microsoft Research), Microsoft Research, Open Source Community)
Open Source Code, Documentaiton, User Guide, Code Examples — for Data Scientists
- Aequitas (University of Chicago Center for Data Science and Public Policy)
Open Source Code, Web Audit Tool, Example, Documentation — for ML Developers, Analysts, Policymakers
- Ethics & Algorithms Toolkit (Johns Hopkins Center for Government Excellence (GovEx), City and County of San Francisco, Harvard DataSmart, Data Community DC)
Guide, Worksheets — for Government Leaders, Stakeholders, Data Analysts, Information Technology Professionals, Vendor Representatives - Consequence Scanning Kit (Dot Everyone)
Manual, Exercises — for Team Members, User Advocates, Tech and Business Specialists, Business or External Stakeholders - AI Ethics Cards (IDEO)
Cards — for Designers - What If Tool (People + AI Research Team (Google))
Open Source Code, Tutorials, Documentation, Examples — for Data Scientists - Digital Impact Toolkit (Stanford Digital Civil Society Lab)
Checklists, Worksheets, Reading Materials — for Civil Society Organizations - Deon Ethics Checklist (DrivenData)
Checklist, Open Source Code, Documentation — for Developers - Design Ethically Toolkit (Kat Zhou)
Exercises, Worksheets — for Designers - Lime (Macro Ribeiro, Sameer Singh, Carlos Guestrin (University of Washington); Open Source Community)
Open Source Code, Documentation — for Data Scientists - Weights and Biases (Weights and Biases)
SaaS product, Articles — for Developers - Responsible AI in Consumer Enterprise (integrate.ai)
Guide, Framework — for Organizations, Executive Leadership, Implementation teams - Algorithmic Equity Toolkit (AEKit) (ACLU of Washington, Critical Platform Studies Group, Tech Fairness Coalition)
Activities — for Community Groups - LinkedIn Fairness Toolkit (LiFT) (LinkedIn)
Open Source Code, Documentation, Blog — for Machine Learning Developers - Audit AI (Pymetrics)
Open Source Code, Documentation, Examples — for Data Scientists - TensorFlow Fairness Indicators (Google)
Open Source Code, Documentation, Examples — for “Teams” - Judgment Call (Microsoft Research)
Cards, Activities — for Technology builders, managers, designers - SageMaker Clarify (Amazon)
Proprietary Code, Documentation, Example — for “AWS customers” - NLP CheckList (Marco Tulio Ribeiro (Microsoft Research), Tongshuang Wu (University of Washington), Carlos Guestrin (University of Washington), Smaeer Singh (UC Irvine))
Open Source Code, Documentation, Examples — for Team - HAX Workbook and Playbook (Microsoft Research)
Guide, Workbook/Worksheets, Examples, Guidelines — for UX, AI, project management, and engineering teams - Community Jury (Microsoft)
Activity — for Product Team - Harms Modeling (Microsoft)
Activity — for Technology Builders - Algorithmic Accountability Policy Toolkit (AI Now)
PDF Guide — for Legal and Policy Advocates