Written by Grace Abuhamad.
The AI Policy Congress benefitted from a unique international perspective as members of the Organisation for Economic Co-operation and Development (OECD) Artificial Intelligence Expert Group (AIGO) participated in the Congress and held their third meeting on draft AI policy principles on the following two days. MIT facilitated international engagement on AI policy principles, which the AIGO plans to provide to OECD governments for consideration this summer.
To foster this engagement, the Congress included opening remarks from Andrew Wyckoff, Director of the OECD work on Science, Technology, and Innovation. In the afternoon, Anne Carblanc, Head of the Digital Economy Policy division, presented the draft AI policy principles and participated in a panel discussion with AIGO members Nozha Boujemaa, Cyrus Hodes, and Osamu Sudoh, about the policy development process. The discussion was moderated by IPRI’s Technology Policy Director, Taylor Reynolds, himself a former OECD staffer.
What Is the OECD?
Since members of the audience may not have been familiar with the OECD, Wyckoff started off with an introduction to the intergovernmental organization. Originally a group of developed market economies, the OECD had earned the reputation of being “the rich country’s club,” but Wyckoff argued that this is no longer the case as the 36 member countries now include Mexico, Turkey, Chile, and soon, Colombia and Costa Rica. The purpose of the OECD is to share best practices for common problems and facilitate cross-border activities that are fundamental to the modern economy. The recommendations coming out of the OECD are not legally binding, but since membership includes adopting them, they reduce inconsistencies and enhance compatibility and interoperability of rules across countries. The OECD has engaged in Internet policy almost since its inception. As early as 1980, the OECD produced Guidelines on the Protection of Privacy and Transborder Flows of Personal Data (these were updated in 2013). In 2011, the OECD produced the Principles for Internet Policy Making that further preserved the free flow of information.
How Has the OECD Engaged on AI?
The OECD has made AI one of its three central priorities for the organization over the next two years. As Wyckoff and Carblanc noted, there is a sense of urgency among member states to coordinate on AI policy, especially with regard to productivity, jobs and skills, and public trust. This has resulted in three work streams for the OECD: shared AI policy principles, an “Observatory” for policy analysis on AI, and more evidence-based analysis and reporting. These efforts will improve information dissemination and enhance coordination among member states as well as help to avoid a patchwork of country-specific or regional approaches that might sow confusion and an unpredictable policy environment.
Carblanc presented the draft AI policy principles process, which began in May of last year, with the convocation of a multistakeholder group (AIGO) including over 50 experts from governments, business, the technical community, and civil society. The group also includes IPRI’s Daniel J. Weitzner, Taylor Reynolds, and CSAIL PhD student Jonathan Frankle. AIGO’s mandate is to scope principles for public policy and international cooperation to foster trust in, and adoption of, AI. Some of the challenges Carblanc noted for the principles process include:
- To be specific to AI, a term with many definitions;
- To be implementable in policies and practices throughout the lifecycle of AI, from its design through to its deployment and use; and
- To be sufficiently flexible to stand the test of time.
AIGO has identified 11 principles, which they have divided into two groups: general principles for AI stewardship and principles for AI policymaking. Each principle is a high-level commitment to a certain value or aspiration related to AI, followed by a non-exhaustive list of implementation options. This way, the principles are actionable for member countries.
Carblanc noted that much of the work and process of AIGO will not necessarily be featured in the final AI policy principles document, which is meant to be simple and high-level. She suggested that these efforts may feed into the Observatory that the OECD will launch in the next few months.
How Does an International Organization Build Consensus on Issues Unresolved at the National-Level?
Carblanc’s presentation of the policy principles development process was followed by a panel discussion on the interplay between national and international policy efforts. Boujemaa, who is the Research Director at the National Institute for Research in Computer Science and Automation (INRIA) in France, described how France gains momentum from efforts such as the OECD process to fund AI research and development. Sudoh, a Professor at the University of Tokyo, advised the Japanese government as it developed its own set of draft human-centered AI principles. As Japan prepares to host the G20 this year, its coordination nationally and within different international organizations would have ripple effects on international AI policy. Hodes, Advisor to the UAE Minister for Artificial Intelligence, agreed with the coordination and investment effects of international principles process. He suggested that the OECD principles may have more of an effect than other international processes because they are developed by like-minded countries and other countries are watching.
How Can MIT Contribute to an International Process?
Panelists reinforced the importance of capacity-building, nationally and internationally, which is a priority of the OECD’s engagement on AI. While still under development, the Observatory will include a public-facing interactive platform with AI policy analyses, national AI strategies, and associated measurements. Wyckoff called on those present at the AI Policy Congress to engage and become Observatory partners. This week, MIT is formally launching its Stephen A. Schwarzman College of Computing which could become a trusted collaboration partner for international AI policy.
Learn More About the MIT AI Policy Congress
- AI Policy Congress – Part 1 Governance Challenges
- AI Policy Congress – Part 2 Democratizing AI through Transparency and Education
- AI Policy Congress – Part 3 Healthcare
- AI Policy Congress – Part 4 Criminal Justice & Fairness
- AI Policy Congress – Part 5 Transportation & Safety
- AI Policy Congress – Part 6 Manufacturing & Labor