Table of Contents
In order for companies and society to utilize AI with confidence and achieve sustainable growth, the Association will discuss AI governance among various players and make policy recommendations based on such knowledge.
Background of the AI Governance Association
The popularity of generative AI has brought about major changes in the business world, and its range of applications is expanding like never before. Many companies have begun to position "AI utilization" as part of their growth strategies, creating a movement that has been called the "fourth AI boom.
At the same time, however, AI-specific vulnerabilities, or "AI risks," which are qualitatively and quantitatively different from those of conventional software, have also come to the public's attention: problems such as AI-induced halting, harmful output, and generation of false information are factors that can pose serious risks to society and businesses. As a result, there is a growing demand for "AI governance" to manage and counter AI risks for companies implementing AI in society.
In the EU and the U.S., there is a growing movement to require AI model developers and AI service providers to manage risks in the development and operation process of AI models, including those of third parties. AI model developers and AI service providers to manage risks in the process of developing and operating AI models, including third parties.
The Japanese government, as the chair of the G7, is also promoting various initiatives, including a discussion of "external audits" to ensure transparency in the guideline skeleton for business operators released in September. AI governance is not confined to specific departments or individual companies, but is being discussed as a collaborative effort among various players.
awareness of the issues
With the aforementioned challenge of promoting the sound use of AI, many players continue to search for how AI governance should be put in place to promote its use. To enable companies to take on the challenge of AI utilization without fear of risk, it is necessary to share AI governance practices across industries and create a state where each company can implement AI with confidence based on a certain common understanding.
In addition, the government has already provided certain AI governance guidelines, but there is no visibility into whether individual companies are implementing these guidelines, and companies that have developed appropriate AI governance are not yet evaluated by the market. The absence of a framework for evaluating AI governance is potentially a hindrance to the healthy use of AI by companies.
In general, establishing standards around AI governance, clarifying the roles of each player, and forming an ecosystem for sound AI utilization are important issues for AI utilization in Japan.
Outline of the Association and its activities
To address these challenges, we are launching the AI Governance Association with companies from across the industry.
We have also been holding study groups of industry volunteers to discuss what kind of AI governance is needed for sustainable development of the industry. Now, as the G7 is moving forward with concrete steps such as the formulation of the Hiroshima AI Process, we are initiating discussions in the form of an open consortium in order to invite more diverse collaborators.
Through our discussions, we aim to establish a standard for AI governance that takes into account the latest practices and domestic and international institutional trends, thereby contributing to the individual discussions held by each industry sector.
At present, the Association plans to carry out the following two major activities.
(1) Foster cross-industry discussion and common understanding of AI governance among developers of infrastructure models and AI service providers through study groups, etc.
(2) Study on institutional arrangements for AI and make policy recommendations to the government
(1) Foster cross-industry discussion and common understanding of AI governance
We will develop a forum for companies challenging AI utilization in Japan to share practices for AI governance and discuss what it should look like. The following activities are currently envisioned
Develop "AI Governance Action Goals" as an industry
Based on multi-stakeholder opinions and domestic and international institutional trends, we will define action goals (values to aim for) in terms of governance that companies should realize in their use of AI. By soliciting corporate commitment to the action goals, we aim to foster a common understanding of the baseline for AI governance in Japan.
Share knowledge of AI governance in research groups, etc.
Member companies and experts will share and accumulate knowledge by establishing study groups and other forums to introduce and discuss practices implemented in AI governance practice to realize the above action goals. Based on such knowledge, we will conduct activities such as policy recommendation as described in (2) above.
(2) Study on institutional arrangements for AI and make policy recommendations to the government
The Association will gather and disseminate industry opinions on the rapidly changing policies related to AI governance, as well as study and make proposals on specific policy frameworks. The following activities are currently envisioned, but we will continue to address a variety of agenda items as the situation changes.
Public comments on new policies, etc.
Currently, the government, led by the AI Strategic Council, is compiling discussions on the Hiroshima AI Process and discussing the formulation of "Guidelines for New AI Business Operators" in Japan. As the industry needs to gather and disseminate its opinions on these developments as necessary, the Association will participate in the discussions by submitting public comments and other means.
Study and proposal of institutional framework for certification system, etc.
As mentioned above, there is a great demand for a framework to evaluate companies with a certain level of AI governance. Therefore, in the mid- to long-term, in order to spread the realization of AI governance goals indicated in policy to private companies, we will also study and propose a framework such as a certification system to visualize companies that have established a certain level of appropriate AI governance.
As shown below, a number of industry leaders have already expressed support for the concept of an AI governance certification system, sharing the recognition that a framework is needed to clarify social responsibility and provide visibility to companies that are fulfilling it.
Issues surrounding AI risks are piling up, but we cannot let that stop us from making progress. Through the establishment of this association, we will strive to curb its negative impact and promote the use of AI in Japan by establishing a structure for leading players in Japan to discuss AI governance.
We are looking for companies to join our association in the future. For more information, please contact us at the URL below.
List of companies endorsing the framework of the "AI Governance Certification System" (in alphabetical order)
Comments from concerned parties
Jun Okada, Attorney at Law, Mori Hamada & Matsumoto
In light of the development of generative AI, discussions on a flexible and flexible risk response framework that does not inhibit innovation are in full swing, as more transparency and reliability in AI are demanded. There is a limit to the government's ability to create uniform detailed rules in a rapidly changing field, and in the summary of issues released by the AI Strategy Council in May of this year, it was suggested that reference be made to third-party certification systems and auditing systems, etc. If an appropriate framework can be established to visualize the status of AI governance to a reasonable extent, while being aware of international interoperability, it will contribute to business predictability and competitiveness, in addition to protecting rights and interests. We hope that this forum will serve as a forum for deepening cross-sectoral discussions, transcending differences in industry and position.
Takafumi Ochiai, Attorney at Law, Co-Founder and Representative Director, Smart Governance Co.
AI expands human potential, and promoting AI implementation is an important issue. Even though AI has become accessible through generative AI, it is not clear what reliable AI use looks like, and hesitancy to use AI has occurred. Given the changing social and technological landscape, a voluntary certification system that incorporates an agile governance frame and that involves experts is expected to lead to trust in AI use and can serve as a trump card to promote AI implementation.
Yuko Harayama, Professor Emeritus, Tohoku University
As AI not only improves the efficiency of production activities but also permeates our daily lives, there is a growing awareness of the need for a governance framework that takes into account everything from research and development to social implementation, due to its effectiveness as well as its significant impact on society. An internationally compatible certification system is a candidate for such a framework, and the time has come for comprehensive and concrete discussions from design to implementation, with high expectations for the effective use of a multi-stakeholder approach.
Masashi Ikutame, Managing Executive Officer, CDO, Group Digital Strategy General Manager, Tokio Marine Holdings, Inc.
AI technology is advancing rapidly, and the recent emergence of generative AI is further expanding its potential for use. We support the introduction of the AI Governance Certification System because it is expected to improve the reliability and safety of AI applications, encourage the use of AI technology in business, and accelerate the development of the Japanese economy.
Yukito Oshiba, Executive Director, AI Governance Association and Co-Founder, Robust Intelligence
With the rapid progress of AI utilization, including generative AI, society is undergoing a period of transformation. In this context, we believe that the ability to properly manage "AI risks" such as security, ethics, and quality will be the dividing line between success and failure in the use of AI in Japanese society, and have established the "AI Governance Association". In the U.S. and Europe, implementation of AI governance by industry is progressing. However, no other country is currently striking a balance between the risks and rewards of AI as Japan is. Through this association, we will form an AI governance ecosystem in Japan that can provide guidance to the world through the collaboration of various players, including large corporations, start-ups, and public organizations.