Japan's Digital Agency Tests Generative AI for Governance

The Digital Agency of Japan has announced a large-scale test of its generative AI platform, "Gennai," starting in May. The initiative will involve approximately 180,000 staff across various government agencies and ministries, focusing on the impacts of generative AI on administrative tasks and reforms to work processes. This trial precedes the planned full implementation of the system in fiscal 2027, with the agency also exploring AI solutions to aid in preparing parliamentary statements and remarks.

Significantly, the agency selected seven domestically developed large language models (LLMs) for this trial, emphasizing support for local companies and research institutions in AI development. Upon evaluating the LLMs during the test, the agency will consider the procurement of the highest-performing model by fiscal 2027. This move represents a strategic shift towards enhancing national AI capabilities and limits dependency on foreign technology in public sector applications.