Ethical AI
@index
For reference, see definitions of AI .
Also relevant are the "Ethics 4 AI" lecture notes .
On ethical AI principles and guidelines
- EU Ethics Guidelines for thrustworthy AI
- Verbeek's begeleidingsethiek
- Algorithm Watch - AI Ethics Guidelines Global Inventory
- AI ethics as a way to avoid legislation
- Hagendorff - Ethics of AI Ethics: Evaluation of Guideline
- Bietti-From_Ethics_Washing_to_Ethics_Bashing
- Canca-Operationalizing_AI_Ethics_Principles
Corporate Ethical Principles statements
Many companies nowadays formulate AI principles. They are often brief and very general, with no indication how they are operationalized.
- Ethische benadering AI bij NS
- Microsoft AI principles
- Google AI principles
- TODO IBM -> Cutler, A., Pribić, M., Humphrey, L. (2018). Everyday ethics for artificial intelligence: A practical guide for designers & developers. Retrieved February 04, 2019. https ://www.ibm.com/watson/asset s/duo/pdf/everydayethics.pdf: 1–18.
- TODO Deepmind
- TODO OpenAI -> https://openai.com/charter/
- Notable exemptions are Facebook and Twitter.
- Update 2021: Introducing our Responsible Machine Learning Initiative.
TODO:
- SAP AI-principes
- Gartner Symposium 2019: "AI and Ethics: What you need today and tomorrow"
General notes on AI Ethics
- What is AI ethics?
- Is AI ethics business ethics?
- AI ethics must partially become microethics
- Ethical codes cannot replace ethical reasoning
- Can we operationalize AI ethics?
- AI as arms race
- Algorithmic choice
- AI ethics of objective function
- Digital ethics shopping
- TODO read Floridi paper and add ref
- MVEP: Minimum Viable Ethical Product
- the Collingridge dilemma
- Neutrality Thesis of Technology
Ethics of algorithms
Machine morality
Meaningful human control
See index Meaningful Human Control
Transparency and explainability
See index XAI explainable AI .
Blogs
- Interpretable machine learning (towardsdatascience.com)
- Walks through the Kaggle explainability crash course linked above
- Push limits of explainability with SHAP
- Black-Box models are actually more explainable than a Logistic Regression
- BOUDE CLAIM! Gaat ook over SHAP
- TODO Great overview on Neural Feature Visualization
Books
- Christop Molnar (2021) - Interpretable Machine Learning: A Guide for Making Black Box Models Explainable
- Also covers Shapley values and LIME
- "All interpretation methods are explained in depth and discussed critically. How do they work under the hood? What are their strengths and weaknesses? How can their outputs be interpreted? This book will enable you to select and correctly apply the interpretation method that is most suitable for your machine learning project."
Fairness and Bias
Blogs:
Diversity
Accountability and responsibility
- Matthias - Responsibility Gap Learning Automata
- Raji-Closing_the_AI_accountability_gap
- Describes SMACTR accountability methodology
AI Ethics Tools
- Morley et al - From What to How: An Initial Review of Publicly Available AI Ethics Tools, Methods and Research to Translate Principles into Practices
- For an interactive exploration of AI principles, see TOOLBOX: Dynamics of AI Principles.
Relevant notes
- index_what_is_explanation
- E.g. in interpretable models, are the offered explanations contrastive?
- Why P rather than Q?
- Zie ook EU ethics guidelines for trustwrothy AI: wat is een "explanation" daar?
- Misschien is dit ook een punt waar ik het NS beleid kan concretiseren
- Jan_Broersen-ACAIS_Responsible_AI
- PI-Hoofnagle_e.a.-EU_GDPR.md
- Russell-Human_Compatible.md
- neutrality_thesis_artefacts
- Dual-use discussion
- Black Box Society (concept by F. Pasquale)
- Moral backpropagation
Idee: op basis van wat ik lees een leesgroepje starten over ethische AI
Relevante NS documenten:
- Toekomstvisie 2020-2025
- Ambitie NS
- Wij zijn NS, Gedragscode voor NS Medewerkers
- NS-Principes voor omgang met data en informatie
- Behandelingsvoorstel Raad van Bestuur – Visie Artificial Intelligence (AI) voor NS, A. Chernilovskaya, augustus 2019
Tutorials
Project Ideas
Explanatory power and reliability of local explanations
Idea popped in my head when reading Mittelstadt_Russell_Wachter-Explaining_Explanations_in_AI :
Also, this makes me think of an interesting research project: How reliable are local approximations (i.e. 'explanations') in terms of the global system behavior? Without guarantees here, what explanatory power do these explanations really have? Do measures already exist that express the reliability of the explanations?
Reading list
- Algorithmic Impact Assessment!
Reisman, D., Schultz, J., Crawford, K., & Whittaker, M. (2018). Algorithmic impact assessments: A practical framework for public agency accountability. Retrieved from AINow website: https ://ainowinstitute.org/aiareport2018.pdf.
Saltz, J. S., & Dewar, N. (2019). Data science ethical considerations: A systematic literature review and proposed project framework. Ethics and Information Technology. https ://doi.org/10.1007/s10676-019-09502-5.
-
D. Leslie, Understanding artificial intelligence ethics and safety, 2019, doi: 10.5281/zenodo.3240529 .
- From Alan Turing institute
- Arrieta from above refers to this paper for a four-step methodology for implementating responsible AI
- What I like about this, is that it explicitly takes context and the targeted audience into account, instead of doing "plug and play" AI
-
AI4People framework
Floridi, L., Cowls, J., Beltrametti, M., Chatila, R., Chazerand, P., Dignum, V., et al. (2018). AI4People— an ethical framework for a good AI society: Opportunities, risks, principles, and recommendations. Minds and Machines, 28(4), 689–707. https ://doi.org/10.1007/s1102 3-018-9482-5.
- Seven checkpoints for transparency: Koene et al., 2019
Koene, A., Clifton, C., Hatada, Y., Webb, H., & Richardson, R. (2019). A governance framework for algorithmic accountability and transparency (Study No. PE 624.262) Panel for the Future of Science and Technology, Scientific Foresight Unit (STOA), European Parliamentary Research Service.
- On the usefulness of explanations of machine learning models and predictions: Mittelstadt, Russell and Wachter (2019)
Mittelstadt, B., Russell, C., & Wachter, S. (2019). Explaining explanations in AI. Proceedings of the conference on fairness, accountability, and transparency - FAT* ‘19, 279–288. https://doi.org/10.1145/3287560.3287574
- Challenges caused by opening the black box: Wachter, Mittelstadt and Russell (2017)
Wachter, S., Mittelstadt, B., & Russell, C. (2017). Counterfactual explanations without opening the black box: Automated decisions and the GDPR. Harvard Journal of Law & Technology, 31(2). https://jolt.law.harvard.edu/assets/articlePDFs/v31/Counterfactual-Explanations- without-Opening-the-Black-Box-Sandra-Wachter-et-al.pdf
- GDPR (already partially read)
Wachter, S., Mittelstadt, B., & Floridi, L. (2017). Why a Right to Explanation of Automated Decision-Making Does Not Exist in the General Data Protection Regulation. International Data Privacy Law, 7(2), 76–99. https://doi.org/10.1093/idpl/ipx005
- de Laat 2018: perverse effects of disclosure.
de Laat, P. B. (2018). Algorithmic decision-making based on machine learning from Big Data: Can transparency restore accountability? Philosophy & technology, 31(4), 525–541. https://doi.org/10.1007/s13347-017-0293-z
- Does the GDPR "right to an explanation" lead to a "transparency fallacy"? (Edwards & Veale, 2017)
Edwards, L., & Veale, M. (2017). Slave to the algorithm: Why a right to an explanation is probably not the remedy you are looking for. Duke Law & Technology Review, 16(1), 18–84. https://scholarship.law.duke.edu/dltr/vol16/iss1/2
- DONE Arrieta, Alejandro Barredo, et al. “Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI.” Information Fusion 58 (2020): 82–115.
- Very useful for conceptual distinctions in XAI
DONE Lipton - Mythos of Model Interpretability
DONE From Floridi, L. (2019c). Translating principles into practices of digital ethics: Five risks of being unethical. Philosophy & Technology. https ://doi.org/10.1007/s1334 7-019-00354 -x :