Nevertheless, the Asian nation is currently focusing on implementing \u201csoft\u201d as opposed to \u201chard\u201d laws such as the EU has done with generative AI, in order to ensure that innovation by AI is not impeded. This has been achieved by building a framework for governance and guidelines instead of imposing obligations through laws and regulations. As such, the private sector has been left to conduct the governance voluntarily. An example of such a guideline is the \u201cGovernance Guidelines for Implementation of AI Principles\u201d published by the Ministry of Economy, Trade and Industry (METI) on 9 July 2021 (as amended in Version 1.1 of 28 January 2022).<\/p>\n
However, calls have been made by Japan\u2019s Liberal Democratic Party to develop a new national strategy and to review previous measures as soon as possible, with regard to the Japanese government\u2019s existing AI strategy.<\/p>\n
Challenges that generative AI poses and obstacles to full implementation in Japan<\/h2>\n
![\"Japanese](\"https:\/\/www.1stopasia.com\/blog\/wp-content\/uploads\/2023\/09\/The-Growing-Relationship-Between-AI-and-the-Japanese-Language-300x178.jpg\")
When it comes to the challenges posed by AI and the obstacles that Japan faces with regard to this new development, it\u2019s important to consider what the country has to deal with in order to ensure greater transparency, fairness, and respect and upholding of human rights in relation to implementing generative AI on a wider scale.<\/p>\n
Among the first challenges is the limited availability of a robust community of software engineers<\/strong> to develop the necessary infrastructure and applications. For example, Japan is expected to face a deficit of 789,000 software engineers by 2030.<\/p>\n Apart from software developers, there is also the challenge of hardware<\/strong>, as LLMs need to be trained using AI supercomputers. Currently, no private company in Japan possesses its own machine with the requisite capabilities. The alternative, of course, is to use government-controlled supercomputers in Japan\u2019s pursuit of LLMs.<\/p>\n Furthermore, there is the challenge of ensuring accurate translations<\/strong> of the Japanese language itself. Japanese is a language in which sentences sometimes lack a subject and, as such, natural translations into English are often difficult to achieve. Literature, for instance, is one of the most difficult tasks for generative AI to translate, because it requires a translation of the nuances of a story in an eloquent manner and it requires a more in-depth understanding of the time period being described, which is something generative AI does not yet have the capacity of doing.<\/p>\n In addition to this, there are different prompts<\/strong> that can be given for the same task, producing different results.<\/p>\n Another issue is that generative AI tools can be weak in practical and detailed areas<\/strong>, such as explaining grammar and spelling and doing arithmetic.<\/p>\n Finally, AI chatbots are known for imagining facts and making reasoning errors<\/strong>. LLMs have been accused of being able to \u201challucinate\u201d and present inaccurate information as fact, which many have warned about.<\/p>\n LLM development, and especially Japanese LLM require a lot more work for more accurate AI translation and more accurate and factually correct output. Although there are some legislative steps being taken towards a \u201csoft\u201d regulation of the industry, much remains unknown as to how generative AI will be regulated in Japan. What is clear is that AI for Japanese translation is essential if Japan wants to maintain its status as a high technologically developed country and as competition between the United States and the EU continues.<\/p>\nFinal thoughts<\/h2>\n