IGF 2019 OF #39
Artificial Intelligence – from Principles to Practice

Description

The transformation of society, government and industry being driven by AI systems requires coordinated and forward-looking public policy frameworks that are informed by government, industry, policy, technical experts and the public, to shape human-centric development and deployment of AI. This Open Forum will inform participants about the OECD Recommendation on Artificial Intelligence and seek feedback on priority areas for AI public policy and ideas for multi-stakeholder co-operation.

The OECD’s Committee on Digital Economy Policy agreed to form an AI expert group (AIGO) in May 2018, which completed its recommendations to the OECD in Dubai in February 2019. The OECD’s Committee on Digital Economy Policy then built on the recommendations to develop the first intergovernmental Recommendation for AI in March 2019, that is expected to be adopted at the annual OECD Ministerial Council Meeting in May 2019.

The Recommendation promotes human-centric AI that fosters innovation and trust. Complementing existing OECD standards in areas such as privacy, digital security risk management, and responsible business conduct, the Recommendation focuses on the features specific to AI and sets a standard that is implementable and flexible, so as to stand the test of time in a rapidly evolving field.

The Recommendation identifies five complementary value-based principles for the responsible stewardship of trustworthy AI that are relevant to all stakeholders: inclusive growth, sustainable development and well-being; human-centred values and fairness; transparency and explainability; robustness, security and safety; and accountability. It further calls on AI actors to promote and implement these principles according to their roles.

In addition, the Recommendation provides five recommendations to policy makers pertaining to national policies and international co-operation for trustworthy AI, namely: investing in AI research and development; fostering a digital ecosystem for AI; shaping an enabling policy environment for AI; building human capacity and preparing for labour market transformation; and international co-operation for trustworthy AI.

The OECD has started to move from principles to implementation with the second leg of our work on AI: the AI Policy Observatory to be launched in 2019. Through the Observatory, the OECD is working with a wide spectrum of partners from governments, industry, policy and technical experts and academia. The Observatory is a multidisciplinary, evidence-based and multi-stakeholder centre for policy-relevant evidence collection, debate and guidance for governments, while providing external partners with a single window onto policy-relevant activities and research on AI from across the OECD (more information on: http://oe.cd/ai)

Content of the Session

The first part of the session will focus on presenting and discussing the content of the OECD Recommendation on Artificial Intelligence and its role in helping to shape an international AI policy framework. Following an introduction to the Recommendation, Governments from several countries will provide their perspectives on its content.

In the second part of the panel, partner IGOs and technical, business and civil society representatives will be invited to discuss priorities to help policy makers move from principles to practice and priorities for the AI Policy Observatory. Interventions will build on the advanced initiatives underway including the IEEE’s Initiative on the Ethical Design of Autonomous and Intelligent Systems, the Partnership on AI to Benefit People and Society, the EC and UNESCO’s work on AI.

Interventions

1) The international AI policy framework and the role of the OECD principles for artificial intelligence:

Ms. Audrey Plonk, Head of Digital Economy Policy Division, OECD, will welcome speakers and participants to the OECD Open Forum on AI along with Mr. Yoichi Iida, incoming CDEP Chair, Japanese Ministry of Internal affairs and Communications (MIC-Japan), who will moderate the session. 

Ms. Karine Perset, Administrator – AI Policy, Digital Economy Policy Division, OECD, will present the process to develop the AI principles and next steps for the OECD to help implement the Principles (5 minutes). 

Ms. Makiko Yamada, Vice Minister, MIC-Japan, will presenta the origin of the AI Principles at the G7 in Takamatsu and the significance of the principles including in the G20 process (5 minutes).

2) Priorities for international and multi-stakeholder cooperation in moving from principles to practice, including through the AI Policy Observatory 

Mr. Rob Strayer, Deputy Assistant Secretary of State for Cyber and International Communications and Information Policy, will provide the US Government perspective on the principles and the importance of helping countries to implement the principles, in particular through the OECD AI Policy Observatory (6-7 minutes).

Ms. Carolyn N’Guyen, Director of Technology Policy, Microsoft, will provide business perspective on priorities to implement the AI Principles (3-4 minutes).

Mr. Mina Hanna, Co-Chair of the "Policy Committee of the IEEE Standards Association's Global Initiative on Ethics of Autonomous and Intelligent Systems", will provide technologists' perspective on priorities to implement the AI Principles (3-4 minutes).

Ms. Valeria Milanes, Executive Director of ADC (Association for Civil Rights) and CSISAC Steering Committee member, will introduce a civil society perspective on priorities for AI policy and implementing th OECD Principles (3-4 minutes).

Ms. Sasha Rubel, Programme Specialist, Knowledge Societies Division, Communication and Information Sector, UNESCO, will introduce UNESCO's perspective on priorities for AI policy and the linkages between UNESCO's work and the OECD's (3-4 minutes). 

Ms. Katarzyna Gorgol, Adviser, Digital Affairs and Telecommunication, Delegation of the European Union to the United Nations, will introduce the European Commission's perspective on priorities for AI policy and implementing the OECD Principles and EC's ethical guidelines (3-4 minutes). 

3) Discussants:

Mr. Charles Chew, IMDA (Singapore) will intervene on key AI governance developments in Singapore, Singapore’s participation in AIGO and in G20 (2 minutes).

Ms. Xiao Zhang, from CNNIC/CAC (China) will intervene on China’s priorities for AI development and on the G20 AI Principles (2 minutes).

4) Discussion: The Chair will moderate a short discussion and if possible take 1 or 2 questions from the floor (20 minutes). 

Diversity: This open forum is designed to provide diverse perspectives. It is balanced first in terms of stakeholder groups, with intergovernmental organisation representation through the OECD, UNESCO and the European Commission; private sector representation through Microsoft; technical community representation through the IEEE, and civil society representation through CSISAC. In terms of gender, the workshop will be balanced with the participation of at least three women. Participation of speakers from Asia, North America, Europe will ensure that the panel is geographically diverse.

Organizers

OECD

Speakers

• Mr. Yoichi Ida, incoming CDEP Chair, Japanese Ministry of Internal affairs and Communications (MIC-Japan). 

• Ms. Karine Perset, Administrator – AI Policy, Digital Economy Policy Division, OECD. 

• Ms. Makiko Yamada, Vice-Minister, MIC-Japan.

• Mr. Rob Strayer, Deputy Assistant Secretary of State for Cyber and International Communications and Information Policy, United States.  

• Ms. Carolyn N’Guyen, Director of Technology Policy, Microsoft.  

• Mr. Mina Hanna, Co-Chair of the "Policy Committee of the IEEE Standards Association’s Global Initiative on Ethics of Autonomous and Intelligent Systems". 

• Ms. Valeria Milanes, Executive Director of ADC (Association for Civil Rights) and CSISAC Steering Committee member.  

• Ms. Sasha Rubel, Programme Specialist, Knowledge Societies Division, Communication and Information Sector, UNESCO.  

• Mr. Katarzyna Gorgol, Adviser, Digital Affairs and Telecommunication, Delegation of the European Union to the United Nations. 

Onsite Moderator: Yoichi Iida

Online Moderator: Nobu Nishigata

Rapporteur: Karine Perset

Online Participation:

Remote participation will be facilitated by the remote moderator who will frequently communicate with the remote participants throughout the session to ensure their views/questions are reflected. The workshop will be promoted in advance and during the IGF on the OECD websites and via social media, through the hashtag #IGFOECD.

Discussion facilitation:

The moderator will set the stage by providing context for the workshop, will make sure all the different perspectives are represented throughout the discussions and about halfway through the session will seek input and questions from the floor as well as from remote participants.

Online Moderator

Nobu Nishigata

SDGs

GOAL 9: Industry, Innovation and Infrastructure
GOAL 10: Reduced Inequalities
GOAL 16: Peace, Justice and Strong Institutions
GOAL 17: Partnerships for the Goals

1. Key Policy Questions and Expectations

The OECD’s AI Principles articulate five values-based principles (Inclusive growth, sustainable development and well-being; Human-centred values and fairness; Transparency and explainability; Robustness, security and safety; and Accountability) and five recommendations for policy makers (Investing in AI research and development; Fostering a digital ecosystem for AI; Shaping an enabling policy environment for AI; Building human capacity and preparing for labour market transformation; and International co-operation for trustworthy AI).

The session planned to discuss priorities in the implementation of OECD’s AI Principles from various perspectives, including that of governments, business, the technical community, civil society and intergovernmental organisations. The session also planned to discuss the role of –and priorities for– the OECD’s AI Policy Observatory (OECD.AI), which is being developed as a collaborative platform on AI policy. Launching in February 2020, it aims to facilitate knowledge-sharing, measurement and analysis.

2. Summary of Issues Discussed

There was broad support for the AI Principles adopted by the OECD in May 2019. Speakers highlighted the complementarity and consistency between the OECD AI Principles and many other initiatives in Japan, the US, the IEEE, the Public Voice, UNESCO, the European Commission (EC) ethical guidelines, as well as Singapore and the G20 AI principles.

The OECD presented a ‘sneak preview’ of its AI Policy Observatory (OECD.AI) for launch in February 2020, including features such as guidance on the implementation of the OECD AI Principles, analytical resources across policy areas, trends and data and a database of national AI policies, along with country dashboards. There was a broad support of, as well as strong enthusiasm for, the OECD’s work in developing the policy observatory.

Speakers provided perspectives on the implementation of AI Principles and priorities of the work of the Observatory and there was broad agreement among all participants on:

-  the importance of promoting innovation through AI and at the same time putting in place appropriate oversight to ensure human-centric, responsible AI that respects basic human rights including privacy, as well as fairness and accountability and

 -  the importance of context and risk management approaches when implementing high-level principles for AI, for example, explainability may be critical or not depending on the use context.

Participant from national governments emphasised the importance of AI principles as a foreign policy priority, including in Japan and the United States.  EC presented following priorities on AI: 1/ encouraging investment in R&D, 2/ ethical frameworks, 3/ labor markets and improving skills through training.

There was discussion on the dual use nature of AI, that a tool that can be used as an enabler of good and poverty reduction but also for authoritarian purposes. 

3. Policy Recommendations or Suggestions for the Way Forward

There was broad agreement on the need to move from high-level principles to practical implementation. Singapore presented its “Model AI Governance Framework” as an example. There was strong expectation that  the OECD would continue to lead the international policy discussion on AI through the work on AI Policy Obsevatory (OECD.AI) to provide a collaborative platform on AI policy to facilitate knowledge-sharing, measurement and analysis in multi-disciplinary and evidence-based manner with global multi-stakeholder partners. Among others, Microsoft, which working closely with the OECD to provide live data of AI research and demand-supply of AI talents, emphasised the value of the Obsevatory to help evidence-based policy making.

There was also broad consensus on the importance of public-private partnerships and multi-stakoholder approaches to AI policy, acknowledging the role of all stakeholders in the AI lifecycle and the implementation of principles for trustworthy / human-centric AI. The IEEE, Microsoft and others emphasised their engagement in, and support for, multi-stakoholder approaches to AI policy making.

With respect to future contributions, the OECD looks forward to continuing to input into the IGF, including on development regarding the AI Policy Observatory.  

4. Other Initiatives Addressing the Session Issues

The examples provided in the session included: the OECD AI Principles of May 2019 that set the  first inter-govenmental standard for AI-related policy making; the OECD AI Policy Observatory that will constitute a collaborative platform on AI policy to facilitate knowledge-sharing, measurement and analysis; the Japanese government initiatives to lead discussions on AI in the G7 and G20; the Japanese government AI strategy; the US Government AI strategy “AI for American People”; Microsoft’s AI Principles; the IEEE’s Global Initiative on Ethics of Autonomous and Intelligent Systems and development of P-7000 series of technical standards; UNESCO’s report on ethics of AI as well as the work to develop a standard-setting instrument on the ethics of AI; the Public Voice ‘Universal Guidelines on Artificial Intelligence’, as well as the policy and investment recommendations and ethics guidelines for Trustworthy AI developed by the European Commission’s high-level expert group; the Model AI Governance Framework developed by Singapore and initiatives in China such as the guiding principles toward the development of responsible AI.

5. Making Progress for Tackled Issues

Some examples of multi-stakeholder collaboration were presented during the session, including: the multi-stakeholder process of the OECD’s expert group to scope the OECD AI principles, Microsoft’s engagement in the development of the OECD.AI Policy Observatory and its project in Singapore to develop principles for responsible AI in the financial sector, the activities of “The Partnership on AI to Benefit People and Society”, IEEE’s engagement in the policy development in US such as work with “AI Caucus”, the creation of the high-level expert group within the EC (EU-HLEG) and collaboration between the OECD and the European Commission as well as between the OECD and UNESCO. There was broad agreement that such collaborations are key to tackle global issues on AI, and that they should emphasise multi-stakeholder engagement, inter-disciplinarity and global participation. 

6. Estimated Participation

About 150 participants onsite, of which about half were women. We could not identify online participants from the session site.   

7. Reflection to Gender Issues

The discussion on gender and AI systems was led by Ms. Sasha Rubel from UNESCO, who explained how: (1) AI algorithms could embed gender bias due to uneven representaion of women in the dataset, (2) women’s participation in research, development and use of AI systems should be encouraged. “Women & AI Daring Circle” led by Microsoft is an example to facilitate women’s participation in this field. 

8. Session Outputs

OECD’s work on AI: http://www.oecd.org/going-digital/ai/

OECD’s AI Principles: https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0449

G20 AI Principles:

https://www.mofa.go.jp/policy/economy/g20_summit/osaka19/pdf/documents/en/annex_08.pdf

US Government AI strategy: https://www.whitehouse.gov/ai/

Microsoft AI Principles: https://www.microsoft.com/en-us/ai/our-approach-to-ai

Women & AI Daring Circle, led by Microsoft : http://www.womens-forum.com/initiatives/women-and-AI

IEEE “Ethically Aligned Design”: https://ethicsinaction.ieee.org/

The Public Voice “Universal Guideline for Artificial Intelligence”: https://thepublicvoice.org/ai-universal-guidelines/

European Commission “Policy and investment recommendations for trustworthy AI”: https://ec.europa.eu/digital-single-market/en/news/policy-and-investment-recommendations-trustworthy-artificial-intelligence

European Commission “Ethics guidelines for Trustworthy AI”: https://ec.europa.eu/digital-single-market/en/news/ethics-guidelines-trustworthy-ai

UNESCO “Preliminary study on the Ethics of Artificial Intelligence”:

https://unesdoc.unesco.org/ark:/48223/pf0000367823

UNESCO’s work on AI and ethics: https://en.unesco.org/generalconference/40/results

Singapore Government “Model AI Governance Framework”:

https://www.pdpc.gov.sg/resources/model-ai-gov