See also: ais-ethics-open-groups | ais-ethics-orgs | ais-ethics-standards | codigo-de-etica-brasil (portuguese)
Curated list of standards related to Ethics of Autonomous
and Intelligent Systems (A/IS). AI Ethics, Algorithmic Bias, Ethics of
Automation, Existential Risks, Friendly Artificial Intelligence, Future of
Work, Human Rights, Information Ethics, Machine Ethics, Philosophy of AI,
Privacy, Roboethics, Robot Rights...
"As we keep improving AI, AI researchers start running into standard human problems that people have worked on for decades, centuries, or millenia, and thinking they are the first ones to see them. e.g. defending minority interests, international security, what does "fair" mean." — Joanna J Bryson
In addition to technical standards that focus explicitly in terms related to A/IS Ethics, this document mentions others that, either by being enforced by regional laws and/or have been discussed for much longer in a more specific knowledge area of the A/IS of your interest, are very relevant.
The following documents are used as a direct basis in constitutions and laws throughout all1 the countries on Earth.
Protip: the main link posted here is for the English version, but all these documents have oficial translations for several languages. The most translated one is UDHR with 500+ translations.
Region | Topic | Organization | Document |
---|---|---|---|
Worldwide 1 | Human Rights | United Nations | Universal Declaration of Human Rights |
Worldwide 1 | Human Rights | United Nations | International Covenant on Civil and Political Rights, 1966 |
Worldwide 1 | Human Rights | United Nations | International Covenant on Economic, Social and Cultural Rights, 1965 |
Worldwide 1 2 | Human Rights | United Nations | Convention on the Elimination of All Forms of Discrimination against Women, 1979 |
Worldwide 1 | Human Rights | United Nations | International Convention on the Elimination of All Forms of Racial Discrimination |
The following documents are used by a group of countries and should be considered in A/IS implementations in the respective regions.
Note that under certain circumstances, such as processing and storing personally identifiable information (PII) from citizens of such regions, they may influence A/IS implementations in other regions.
Region | Topic | Organization | Document |
---|---|---|---|
European Union | Data Protection | European Commission | General Data Protection Regulation (GDPR) |
Undefined 3 | Human Rights | Organisation of Islamic Cooperation | Cairo Declaration on Human Rights in Islam, 1990 3 |
The following documents are used by a a country and should be considered in A/IS implementations in the respective regions.
Note that under certain circumstances, such as processing and storing personally identifiable information (PII) from citizens of such regions, they may influence A/IS implementations in other regions.
Region | Topic | Organization | Document |
---|---|---|---|
China | Data Protection | Government of China | 信息安全技术 个人信息安全规范 gdpr-like (via csis.org) |
Brazil | Data Protection | Government of Brazil | Lei Geral de Proteção de Dados Pessoais (LGPDP) gdpr-eq |
Important note: this area will need significative time to catalog standards in a way that someone new could undestand more easily how these where developed.
Region | Topic | Organization | Document |
---|---|---|---|
Worldwide | Multiple | The IEEE Global Initiative | Ethically Aligned Design, First Edition |
This area contains code of ethics, code of conduct and principles. In short, they serve as moral codes. They can restrict what you or your A/IS could do beyond what is enforceable by regional laws.
These documents can serve as a public commitment as a company its employees should act, but can be written in a way that inspires others out of such a company to follow such recommendations.
Note: this section will need some clean up. And maybe even move to a different repository (fititnt, 2019-02-06 06:36 BRST)
Company | Links |
---|---|
Apple | 1) Privacy; 2) Supplier Responsibility; 3) Transparency Report (Government Requests); ... |
1) Facebook Principles; 2) Facebook AI Research | |
Responsible AI Practices | |
IBM | IBM Policies & Principles |
Microsoft | Microsoft AI principles |
TODO: these links need better classification on this document
Organization | Document |
---|---|
Future Of Life | Asilomar AI Principles |
The following references have a very limited application as practical standards and, therefore, are classified here as science fiction. They are explicitly listed because they are popular and as a way to encourage people to understand that there is a difference between them and others.
TL;DR: do not use these references for non-fiction, or at least educate the audience of it's limitations.
Region | Topic | Organization | Document |
---|---|---|---|
"Science fiction" | Roboethics | Isaac Asimov | Three Laws of Robotics (Asimov's Laws) |
"Science fiction" | Roboethics | Osamu Tezuka | Ten Principles of Robot Law / Astro Boy series |
-
wip: These documents are (at least at time of inclusion here) "work in progress" or drafts and do not have at least their first version finished and ready to be used.
-
like-gdpr: These documents seems to be compatible to GDPR.
-
like-gdpr: These documents seems to have similarities to GDPR (but maybe are not equivalent). It can be even more restrictive than GDPR on how to handle personal data.
-
1: These treaties are followed by Member states of the United Nations (e.g near all territories), but some articles of each document may have minor objections. See Multilateral Treaties Deposited with the Secretary-General for full details.
-
2 See reservations on Convention on the Elimination of All Forms of Discrimination against Women ↩
-
3: Except for non-official copies or non-official translations from the Cairo Declaration on Human Rights created on 1990, there are no official links do download and the Organisation of Islamic Cooperation does cite it's endorsement from this very specific document, but has references like the Statute of the OIC Women Development Organization. The Human Rights documents from United Nations with (if any) minor objections seems to be a more reliable source in 2019, even for state members of OIC.
To discuss suggestions and criticisms, you can:
- Open a issue
- Make a Pull request
- Talk via e-mail:
- with the maintainer: Emerson Rocha [email protected].
To the extent possible under law, Emerson Rocha has waived all copyright and related or neighboring rights to this work to Public Domain.