About Us
Chuo University’s Fundamental Attitude toward Generative Artificial Intelligence
June 5, 2023
Chuo University’s Fundamental Attitude toward Generative Artificial Intelligence
Chuo University
President Hisashi Kawai
Research and development of Generative Artificial Intelligence (hereafter “generative AI”) are accelerating rapidly, and we are becoming able to harness various generative AI systems. While traditional systems, typified by databases, provide prepared information gained through high-speed data matching, generative AI systems “learn” from vast amounts of data online, typically on the Internet, and output “new (or apparently new) expression.”
These AI technologies hold great potential to transform the current social structure from the bottom up. Until now, mechanization and robotization of formulaic, repeatable, and continuous work has been taking place. However, we are now in the midst of a debate, how to take advantage of generative AI systems for non-routine work and new tasks in order to develop our society in a sustainable manner, in particular as we face a phase of steep population decline. We can also gain new insights through collaboration between human beings and generative AI, which may lead to a generation of new values in the near future.
As an institution of higher research and education, Chuo University considers it our social responsibility to engage in research and development of generative AI and its usage, including social implementation by paying attention to the following points.
First, as generative AI learns from data on the Internet, which is a part of the real world, we need to be aware that generative AI is vulnerable to the distortions of real society. For example, as long as we cannot exclude hate speech on the Internet, there is the risk of generative AI systems being affected and outputting such expressions. As a result, we should evaluate and treat those generated outputs or expressions from an ethical point of view.
Second, we should always be aware of how algorithms are used to integrate the “learned” information into output/expressed content, and make constant efforts toward its visualization. For instance, as long as antisocial discourses such as hate speech existing in real society cannot be excluded, we should tackle them algorithmically and keep generative AI systems free from those negative impacts. However, the aforementioned requires “someone” to “pre-stipulate” the essence of being “antisocial,” which may pose the risk of eliminating certain values by making them invisible, or of forcing specific values. Therefore, we should consider that the research and development of generative AI systems and their social implementation should be accompanied by researchers' and developers' duty to explain of what values are treated positively or negatively.
Third, while generative AI may contribute to making our lives and societies prosperous, and such prosperity should not be hindered, it is imperative to clarify that output/expression contents are the products of generative AI when using it. For example, if a generative AI user claims that the output/expressed content is their own work, it will not only raise legal issues but also be counter to the autonomy of an individual to decide life and society on one’s own. If the products of generative AI are distributed as “my or our own” decisions, individual liberty based on personal autonomy and its accompanying responsibilities, which are essential parts of our society, will face material risks.
Fourth, it is necessary to use generative AI systems with the recognition that they, at this point, are innovative and developing technology, and that they will never reach "completion" because they originate from information that exists on the Internet. This presents risks of including distortions of real society as mentioned above, misunderstandings of facts, and not taking small but material facts seriously. It is essential to continuously check the output/expression content of generative AI systems when using them.
Fifth, although systems developed on the Internet, including generative AI, are borderless, we must pay attention to large legal and cultural differences among countries and regions regarding research, development, and social implementation of generative AI. For instance, considering the relationships between generative AI and intellectual property rights or privacy rights, many countries have started to enforce their own regulations. There will be cases in which actions that are considered acceptable in Japan will be subject to strong criticism in other countries or regions. It is very important for each of us to have in-depth knowledge of the differences and diversity among cultures and legal systems.
The roles that generative AI plays in the sphere of research education at Chuo University and in Chuo’s contribution to society will surely expand dramatically. Those responsible are kindly asked to always bear in mind the above points and make good use of generative AI in each specialized field.
Please refer to the appendix, “Considerations when using Generative AI in the Chuo University Curriculum,” regarding the use of generative AI systems in Chuo curricula.
