2025-09-20

How Much AI & RPA Do Japanese (National & Local) Government Workers Actually Use?

The employment of Artificial Intelligence (AI) and Robotic Process Automation (RPA)¹ solutions in many industries have demonstrated their potential in greatly reducing or completely removing the need for human intervention or labour². Japan has often been a target of reports and opinion pieces slamming the rather slow adoption and overall low maturity in “digital transformation (DX)”³, which may include the adoption of such technologies in order to keep up with global trends and expectations.

Putting aside the discussion on the reasons why Japan is slow at DX for now, I wanted to have a brief look at how much AI & RPA is being adopted in the public sector in Japan. Despite the Japanese media space being plagued with pieces about slow DX, I admittedly never grasped properly how slow it is.

Ministerial (National) Level: Slow and Steady, But Does It Win the Race?

Officially, Japanese ministries and public sector organisations sporadically utilise AI solutions within their duties. Many cases of AI and RPA usage appear to be mostly a part of experimental runs and pilot studies done by different ministries.

For instance, a few ministries have trialed different AI products to carry out tasks such as optical character recognition, automatic translation, and speech-to-text dictation, with many of them now officially provisioned and adopted by them⁴. Additionally, the National Police Agency (NPA) have carried out pilot studies and experimental usage of AI solutions in a variety of classification tasks, ranging from detecting criminal business transactions to drug distribution channels on social media platforms⁵. Even though there are cases in which government agencies procure and use AI solutions for certain tasks, it appears that there is no unified AI solution used across all institutions at the time of this writing.

With regards to Generative AI, particularly involving Large Language Models (LLMs) from OpenAI or Google, the government so far released guidelines towards the procurement and usage of these solutions, emphasising the importance of raising awareness around cybersecurity and privacy concerns. Furthermore, the Digital Agency (DA) has an initiative called the Priority Plan for the Advancement of a Digital Society and within the explainer document⁶ there are clauses regarding the issues hindering the use of AI within the government, citing:

Based on the “Digital Principles for Structural Reforms” formulated by the Digital Extraordinary Administrative Advisory Committee, we have worked on eliminating analog regulations that hinder the social implementation of AI, digital and other types of technologies, completing revisions of approximately 98% of the seven common types of analog regulations.

The DA, being the flag carrier for DX within the Japanese government, is understandably taking a lot of responsibilities for the development and maintenance of smart systems meant to facilitate civil service. One major aspect of this is the rather bluntly named “Government AI”, which is meant to be used government-wide and by local governments. Government AI appears to be a full suite of AI-powered tools which may include Generative AI capabilities (ran via a sub-component called “gennai”, which is the phonetic expression of “GenAI”)⁷. The aim seems to be to offer a unified experience across different organisations.

While there are more initiatives and case studies to look through, it is safe to say at this point that the national government of Japan is steadily implementing AI and RPA solutions to modernise and facilitate civil services. With that being said, there is still a lot of inconsistencies with the procurement and deployment of different tools, not to mention the existence of roadblocks in the form of archaic regulations that are still in place.

Local (Regional/Municipal) Government Level: Widely Implemented, Poorly Adopted?

There are more details on AI and RPA solutions utilised within local governments. Unlike the national government, where procurement terms may be stricter and more challenging to maintain, local governments seemingly have been adopting AI and RPA capabilities more quickly by taking advantage of private services offered by solutions such as LoGoAI⁸.

With regards to AI, there are many local governments who took advantage of generative AI’s abilities to create promotional material, automate meeting transcriptions and even in some cases develop “low code” applications⁹. Due to the nature of tasks often carried out in local governments, one would assume that the perceived improvements in efficiency and satisfaction would also be high. However, it appears that many governments are still somewhat skeptical of the effects of AI to their workflow, as among the issues listed out, “the effects of implementation are unknown” came out as one of the top three major concerns. The other two also highlight the common issues with generative AI, with them being “lack of knowledgeable staff” and “concerns over the legitimacy of generated content”¹⁰. Despite the rapid proliferation of generative AI, their use cases seem to be still limited.

The story is similar with RPA, with the majority of prefectural and municipal governments having introduced some form of RPA process into their workflows, mostly to automate financial and childcare/healthcare services. One thing to observe though is that while many AI solutions were introduced and tested at a low cost (in some cases even for free by using publicly available interfaces), RPA solutions require more investment as it is often a collection of different solutions packaged into a single service. Cost is seemingly a major concern for many governments, rather than the lack of perceived benefits like generative AI¹¹.

What can be said for both AI and RPA deployments in local governments is that there is a lack of knowledge (and staff) that can enable more meaningful usage of these solutions. This is an issue identified in common, and may not be an easy problem to solve as in Japan, generally policies and resources trickle down from the central government. Therefore unless the national government ramps up the deployment and operation of these advanced technologies, it is unlikely that the country will see any significant advancements in civil service, nor there will be incentives by businesses to develop and offer unique solutions.

Footnotes