OECD member and partner countries officially adopted the first set of intergovernmental principles on artificial intelligence (AI) on 22 May 2019. In doing so, they commit to international standards aimed at ensuring the design of AI systems that are robust, safe, fair and trustworthy.
Ahe 36 OECD member countries, as well as Argentina, Brazil, Colombia, Costa Rica, Peru and Romania, have joined the OECD Principles on Artificial Intelligence during the Annual meeting of the Council at Ministerial levelThis year's theme is "The digital transition for sustainable development". Developed with the help of a group of more than 50 experts from all walks of life - government, academia, business, civil society, international bodies, the technical community and trade unions - they are based on five value-based principles for the responsible deployment of trustworthy AI and five recommendations applicable to public policy and international cooperation. The objective is to assist governments, organizations and individuals to design and operate AI systems that best serve the public interest and to ensure that designers and operators are held accountable for their effective operation.
" Artificial intelligence is revolutionizing our lifestyles and work and offers considerable benefits to our societies and economies. Yet it also raises new challenges, concerns and ethical issues. It is therefore incumbent on governments to ensure that AI systems are designed to respect our values and laws to ensure the safety and privacy of individuals. "said Angel Gurría, OECD Secretary-General. « These Principles will provide a global benchmark for building a trustworthy AI, so that we can make the most of the opportunities it offers, for the benefit of all. ». (View the full speech.)
The Principles on Artificial Intelligence are supported by the European Commission, whose High Level Expert Group has drafted Ethical Guidelines for a Trustworthy RNand will be on the agenda of the next G20 Summit, to be held in Japan. In the coming months, digital policy experts at the OECD will work to develop practical guidance to support their implementation.
Although not legally binding, the OECD Principles in other policy areas have paved the way for the development of international standards and assisted governments in the design of national legislation. For example, the OECD Guidelines for the Protection of PrivacyThe Privacy Act, which sets limits on the collection and use of personal data, underpins many privacy frameworks and laws in the United States, Europe and Asia.
Similarly, the G20 and OECD Principles of Corporate Governance have become an international reference for policy makers, investors, companies and other stakeholders working on institutional and regulatory frameworks governing corporate governance.
In summary, they specify that:
- AI should serve the interests of individuals and the planet by promoting inclusive growth, sustainable development and well-being.
- AI systems should be designed to respect the rule of law, human rights, democratic values and diversity, with appropriate safeguards, such as allowing for human intervention when necessary, in order to move towards a just and equitable society.
- Transparency and responsible disclosure of information related to AI systems should be ensured to ensure that individuals know when they interact with such systems and can challenge the results.
- AI systems should be robust, safe and secure throughout their lifecycle; any associated risks should be assessed and managed on an ongoing basis.
- Organizations and individuals responsible for developing, deploying or operating AI systems should be responsible for their proper functioning in accordance with the above principles.
The OECD recommends that governments should undertake the following actions:
- Facilitate public and private investment in research and development to stimulate innovation in trustworthy AI.
- Promote the development of accessible AI ecosystems, including digital technologies and infrastructures, as well as mechanisms for sharing data and knowledge.
- Build a framework for action that paves the way for the deployment of trusted AI systems.
- Equip individuals with the skills they need in the AI field and ensure a just transition for workers.
- Foster transnational and cross-sectoral cooperation to share information, set standards and collaborate on a responsible approach to trusted AI.
(Source : OECD, May 22, 2019)
