While Japan takes pride in its status as an economic powerhouse, its productivity still lags behind other developed nations. However, amid recent technological advances, the emergence of artificial intelligence (AI) and particularly Large Language Models (LLMs) holds the potential to become a new breakthrough for improving productivity in Japanese companies.

■The reality that full-stack websites can be built even without programming experience.

 This website was created by myself (our company's representative) using a Python-based web framework (Django). However, I am not a Python expert. I have no experience in coding, let alone being seriously involved in system development. Nevertheless, I was able to learn the basics of the development process through conversations with ChatGPT, entrust most of the coding to ChatGPT, and actually build a website.

  I did not use so-called "no-code/low-code" services with reduced programming burden (such as CMS tools exemplified by WordPress). For the frontend, I used a Bootstrap5 template purchased for $100 from an overseas template site, and for the backend, I wrote Python code "intensively" (or more precisely, had ChatGPT write it). I then deployed it to cloud environments provided by the world's top three cloud providers, connected it to databases and storage, and was able to build a rudimentary web application.

　Through this experience, I was reminded once again that having some understanding of programming is critically important for grasping the essence of IT at a deeper level when trying to understand IT business domains. Of course, it would be shallow to think that building a rudimentary website means understanding everything about IT. However, by experiencing the use of VSC (coding software), installing various libraries through command lines, and setting up development environments myself, I feel there has been significant progress within myself that was previously unimaginable.

　This website can be called an "AI native" website. This was previously unthinkable. In this column series, I would like to examine the impact that large language models will have on business and society, drawing from my own surprising experiences.

■IT is built in English.

What I was reminded of again through this experience is the harsh reality that IT is built in English. What I strongly realized again while writing programs is the obvious fact that computer languages are basically derivatives of English. Countries where English is the first language, or countries with native languages structurally similar to English, probably have learning costs for programming (including psychological barriers) that are more than double the advantage compared to countries like Japan that do not have such advantages.

　As can be stated again from this personal experience, global IT business has been led by people from English-speaking regions. Therefore, Japanese companies lacking personnel proficient in English faced the dilemma of always being one step behind when incorporating new technologies and information. However, the emergence of LLMs (Large Language Models) has the potential to significantly change this environment. Since LLMs have the ability to understand and utilize information not only in English but also in other languages, such as Japanese, it is possible to considerably overcome previous disadvantages. With AI supporting multiple languages, information access gaps will narrow, and Japan's delayed DX transformation may accelerate in the future.

■ The possibility of streamlining daily operations through LLMs without changing unique business processes or decision-making processes

I also feel that LLMs have the potential to contribute to overcoming disadvantages rooted in the unique culture and traditions of Japanese companies, in addition to language barriers. Japanese companies tend to value unique business processes and decision-making processes. These processes are deeply rooted in corporate culture and strategy and cannot be easily changed. However, LLMs have the potential to streamline daily operations without changing these processes. As a unique direction for Japan's DX, RPA has attracted attention and been implemented in workplaces for several years, but LLMs have the potential to take a similar approach while delivering incomparably greater impact.

■Launching new businesses or startups using LLMs may be surprisingly difficult

　While LLMs can bring tremendous impact, they will certainly pose significant threats to many people and companies. While the emergence of LLMs strongly raises expectations for new AI startups, the overly powerful capabilities of LLMs may pose threats to many startups. I cannot help but feel keenly aware of the risk that themes that startups would previously stake their company's fate on challenging could be instantly replaced by a single ChatGPT plugin function. Our company will continue to closely monitor LLM trends, which bring various possibilities and threats.