SGRA Kawaraban (Essay) in English

Chen Hongyu “What AI can do and cannot do”

AI (artificial intelligence) is a word that Professor John McCarthy, a computer scientist and cognitive scientist, suggested in 1956. It is defined as “science and technology that can make intellectual machines, especially intellectual computer programs”. At present, research on AI has progressed, and definitions of AI differ depending on scientists and their fields. Automation and streamlining in various fields are facilitated by AI. For example, autonomous driving of automobiles, inspection of defective products, and detection of unauthorized entries on credit cards. Those technologies are fulfilled by the acquisition and study of large quantities of information through the recognition and reading of pictures and voices by computers. As we can use ChatGPT without any professional knowledge, it is now a worldwide topic. It can generate, summarize, and collect information through natural communication, conversation or sentences with other people automatically.


My research field is organic chemistry. As a new current in the development of chemical reactions, the utilization of machine learning and data science is drawing attention. At present, the development of an asymmetric catalyst that can control the three-dimensional chemistry of data science under the trials and errors of researchers is ranked as a challenge in organic synthesis. I have been working at my doctoral course on building and demonstrating methodologies that adopt machine learning to the system design of stereoscopic branched type asymmetric catalysts that supply complex molecules to green. I succeeded in the efficient development of a complicated but new catalyst system and could realize the convenience of AI.


On the other hand, many people began to worry that the development of AI will take away the work of human beings. At present, AI is specialized in specific issues, like the solution of simulated or mathematically modeled issues. However, when singularity (technical singular point: at the time of new AI, which is at the same level as the brains of human beings) will get near in the near future, AI will alternate what only human beings can do, and it is the big change in the living environment of human beings. It is likely that AI will be applied not only to simple tasks like cleaning and delivery, but also to professional fields like medical and financial services.


Then, in a technologically advanced period, what kind of AI works that cannot be alternated are there? In principle, when we reach ‘singularity’, computers can keep the same or higher-level knowledge with human beings. As computers can reduce work and human errors, we can realize cost reduction and efficiency improvements comparing with human beings.

However, there is a possibility that such ‘no-mistake’ perfectness becomes a demerit. For example, take the case of kindergarten teachers. If we set up AI, AI will teach students knowledge or manners, and play with them. However, kindergarten period is important for students’ growth, and it is essential for their formation of characters and sociality that they know emotion and mistake mistakes that ‘human being teachers’ commit unconsciously. If we introduce ‘AI teachers’, it will be difficult for kindergarten students to obtain a suitable lifetime rhythm for their ages.


AI technologies are progressing rapidly. It may take a long time to reach ‘singularity’. But it is an important issue how human beings coexist with AI and complement each other’s.



SGRA Kawaraban 740 in Japanese (Original)



Chen-Hongyu, 2022 Raccoon, Researcher at Otsuka Pharmaceutical Co. Ltd.         



Translated by Kazuo Kawamura

English checked by Sabina Koirala