Transforming Governance: Japan’s Journey with Chat GPT and Generative AI

The launch of ChatGPT, a generative artificial intelligence (AI) tool, has been a watershed moment.^[1] In just two months after its release in November 2022, it attracted over 100 million users, propelling generative AI into the mainstream and expanding accessibility.^[2] This impact extends to Japan, where its influence on society has been equally dynamic and is rapidly finding applications throughout the business industry, education, and now government. In April 2023, the Japanese government initiated five nationwide pilot projects testing the integration of ChatGPT to enhance government efficiency.^[3] This article explores how the Japanese government uses generative AI in governance while also considering arguments for and against its use.

When it comes to generative AI, the Japanese government has traditionally embraced a soft law approach to encourage AI development. Since 2017, the country has been active in crafting frameworks and guidelines on AI regulation. Japan’s approach and strategy in AI governance have not only influenced the Organisation for Economic Co-operation and Development (OECD) recommendations on AI but have also been central in G7 discussions held in May 2023.^[4] In essence, Japan’s perspective on AI centres around avoiding stringent regulation to promote innovation while ensuring user protection.^[5]

In April 2023, the Liberal Democratic Party’s Project Team on the Evolution and Implementation of AIs released the AI White Paper: Japan’s National Strategy in the New Era of AI.^[6] This document is significant as it outlines the integration of AI tools, like ChatGPT, into government operations to enhance administrative functions and support sustainable governance. This document makes the following recommendations. First, it called for a new strategy regarding AI regulation, implementation, and utilisation.^[7] This involves enhancing Japan’s AI development capacity and promoting AI adoption in public services. It also includes formulating policies to support private-sector AI usage and investing in new AI regulatory approaches. The white paper proposed applying these tools within government offices and also to facilitate start-up creation.^[8] A standout emphasis was the immediate need for pilot projects to evaluate AI’s impact and potential in the government’s administrative services. These services include tasks like drafting parliamentary responses, supporting legislative affairs, analysing government statistics, and generating meeting minutes.^[9] Since the publication of this document, Prime Minister Kishida has engaged with OpenAI’s CEO, Sam Altman, to explore OpenAI’s collaboration with the government.^[10] Teaming up with Microsoft, the Japanese government has established data centres in Tokyo and Osaka to strengthen domestic data infrastructure and security, paving the way for nationwide pilot projects. ^[11]

In April, the Yokosuka Municipal Government in Kanagawa Prefecture launched the first pilot project._^[12] ChatGPT was used to create content for city bulletins to promote welfare activities. Additionally, it was integrated with the city’s “LoGo Chat” tool, used by government staff. This integration assisted employees in summarising texts, identifying errors, brainstorming new project concepts, and condensing meeting minutes.^[13] By June, the Yokosuka government had formally adopted ChatGPT as part of its administrative operations.^[14] The trial showed that using ChatGPT for these functions reduced work hours by at least 10 minutes per day. The Yokosuka government emphasised that the incorporation of ChatGPT now automates bureaucratic tasks, allowing government employees to dedicate more effort to delivering personalised administrative services.^[15]

The Fujioka Municipal government in Gunma prefecture has also embraced ChatGPT integration, but with a different approach from Yokosuka.^[16] Here, ChatGPT is employed to draft responses for the assembly. Officials in this region highlighted its value in gathering policy ideas and formulating texts.^[17] Meanwhile, the Tokyo Metropolitan Government is set to follow suit in August. This upcoming integration uses the chatbot for generating documents in question-and-answer format and clerical tasks across all departments.^[18] While ChatGPT integration is being observed nationwide, there are also critics of this new development.

The Tottori Prefectural Government has prohibited employees from using this system, and office computers have been configured to block access to ChatGPT. Governor Shinji Hirai of Tottori emphasised the importance of government staff actively gathering information on their own. He said that if machine-generated words become answers in assemblies and formulate policy, democracy would be abandoned and that decision-making should not be left to machines.^[19] Similarly, officials of Hokkaido prefecture also voiced caution and shared that their administration does not plan to use generative AI.^[20]

Central government agencies and the National Diet have also been part of this debate. In an unprecedented move, Prime Minister Kishida answered questions generated by ChatGPT during a parliamentary session on March 29, which marked the tool’s inaugural use in the National Diet.^[21] This session served as a test of ChatGPT’s performance. Among the proponents of this integration is Economy, Trade, and Industry Minister Yasutoshi Nishimura, who believes that while ChatGPT can be an effective support tool for government operations, human oversight is needed for finalising answers.^[22] Japan’s Digital Agency plans to adopt ChatGPT from fall onwards to generate minutes, draft responses to parliamentary questions, and aid in government statistical analysis.^[23] Other government bodies, like the Ministry of Agriculture, Forestry, and Fisheries, intend to use ChatGPT for updating online application materials. Similarly, the Ministry of Internal Affairs and Communication envisions its use to streamline administrative tasks.^[24]

Although this technology has improved task efficiency and reduced the workload of bureaucrats, concerns persist across various agencies regarding ChatGPT’s integration.^[25] The Cabinet Office and the Ministry of Land, Infrastructure, Transport, and Tourism have indicated they won’t use ChatGPT due to disinformation, cybersecurity, and privacy concerns.^[26] Information generated through generative AI and ChatGPT often has errors, called AI hallucinations, and its information is limited to data only available up until September 2021. Besides concerns surrounding data integrity, there are also ethical issues concerning introducing this tool into legislative sessions.^[27] Lawmakers, civil society, and academics across Japan have voiced concerns that extensive AI integration could end up taking over legislative power. ^[28] Integrating this tool as a way to ask questions and generate answers contradicts the fundamental purpose of government, which is to be a platform for discourse and insight.^[29] Other concerns extend to the business sector, where there have been worries about applicants using generative AI for job applications. Similarly, educational institutions expressed unease about students using AI to complete assignments and exams.^[30] Japanese anime and manga artists have also voiced their concerns surrounding copyright infringements in AI-generated works. ^[31]

One fact remains evident: generative AI tools like ChatGPT are a double-edged sword. They are incredibly powerful and offer remarkable advantages, but they also pose many dangers and concerns. The government cannot ignore such technology, as Japanese businesses and society are already embracing generative AI in various capacities. Consequently, within the realm of governance, the focus has shifted from whether to integrate or use such tools to the more complex questions of how, where, and why they can be effectively employed. Navigating this landscape will be a difficult balancing act, not only for Japan’s government but for administrations worldwide.