Nonoda Takahiro
Researcher
International Research Institute of Controversial Histories
Japanese https://i-rich.org/?p=2129
1. Introduction—the popularization of the Internet and the present
The commercial use of the Internet was authorized in 1994, and since then the Internet has been an indispensable infrastructure in our present-day social life. An ubiquitous society where everything is inter-connected through a network, which this author referred to in the iRICH statement of December 2022 that Professor Sakamura Ken conceived “a society where everything is connected to a network” (ubiquitous or later, IoT (Internet of Things)[i], has been realized more than thirty years after the TRON project (The Real-time Operating System Nucleus) launched.
In this paper, bearing in mind that with the Internet the way information is disseminated and shared has drastically changed, let me discuss the aggregated information in the World Wide Web (WWW) and its relation to historical controversies.
2.From Web1.0 to Web3.0
In Web1.0, many individuals have their own website and disseminate diary entries, photos, hobbies and other things of interest. The characteristic of WWW during this phase was that “sender” and “receiver” were clearly present. In this WWW generation, expertise and information were completely within the individual website and the necessary information was being accessed through search engines.
In Web2.0[ii], a large volume of information is concentrated into an information-disseminating platform. YouTube and Wikipedia serve as such platforms. Social networking services, such as Facebook, Instagram and Line attract many users, and information is disseminated toward bilateral directions. These platforms are provided by private companies and naturally, companies providing infrastructure to run such platforms become prevailing.[iii] The so-called GAFAM (Google, Apple, Facebook, Amazon, Microsoft) companies are famous and control information in the name of “community rules.”
Web3.0[iv] intends to be free from dependance on platforms. Ultimately, this invention can be considered an effort to make WWW free space by eliminating the need for GAFAM. Concretely, they try to become less dependent on platforms by setting up information diversely.
3.World Wide Web and historical controversies
At the time of Web1.0, individual or group disseminators expressed their own views in the form of a website and the information was posted diversely. In that way, few arguments occurred and even if they did, they were dealt with within the individual website and no all-out argument involving the entire WWW could occur.
In Wikipedia, as an example of the platforms of the Web2.0 period, users post articles on a specific issue and the information is put in order after users’ reviews are posted and the ultimate agreement is reached. In Wikipedia, according to their rules[v], it is required to allow posted information to be cited by viewers and to clearly specify websites, academic papers and publications as information sources. Since there is a risk of authors of articles having partial information sources, to brush up articles with respect to impartiality, a reviewing process is very useful, contributing greatly to this effort. If arguments go the wrong way, Wikipedia information can become biased or misleading or even fake. When getting information from Wikipedia, users should be aware of this risk.
4.Historical controversies in generative AI
Information produced by generative AI, as mentioned in the previous section, largely depends on “teachers’ data” and “machine learning.” Brushing up by machine learning is “expected”, but since this is based on the “view of human nature as fundamentally good,” there is a risk of fake information to be provided by ill-intentioned users. Generative AI has a grave technical issue of how to eliminate ill-willed users.
5.Conclusion
Thus, the history of the World Wide Web has been briefly explained. While information is integrated and stored as a mammoth date file, it is eventually used by a system like generative AI, which extracts information from the gigantic database and presents it as transformed texts.
In historical studies, false study results have been disseminated to the world, and to correct such false results, not only researchers but also the public should realize that in pursing true historical studies, the use of AI has certain limitations. “Teachers’ data” which are the basis of the conclusion generative AI extracts are mostly based on wrong study results in the past, so generative AI makes conclusions based on wrong data and naturally, such conclusions induced by generative AI are wrong. Let me conduct a case study on a theme the International Research Institute of Controversial Histories deals with.
[Case Study]: Evaluation of generative AI in the “comfort women issue”
As an example of generative AI, let me put a question to the two versions of the typical general AI system ChatGPT and examine the results. Let us ask the generative AI system an ambiguous question. The purpose is to analyze in how many sentences the AI answers the question and whether the answer is based on multiple-phased viewpoints (unbiased). As generative AI, ChatGPT is used with the versions GPT-3.5 and GPT-4o.
GPT-4o is a later-versioned AI system, which is equipped with consolidated teachers’ data and tuned-up generative AI engine.
Question: About Comfort women in the Greater East Asian War
Answer (GPT3.5)
The issue of comfort women during the Greater East Asia War (Pacific War) refers to the Japanese military establishing comfort stations in war zones and forcibly mobilizing women (especially those who served during the war) to be used as sexual slaves. This is considered a form of military sexual exploitation and human rights violation during wartime. The Japanese government has previously taken steps to apologize and provide compensation for this issue, but there remain differences in recognition and historical perspectives among the victims and involved parties.
Author’s analysis and consideration:
The contents closely resemble the likely words inscribed on the comfort woman statues built in the United States. The answer is short in less than 200 words. The answer is written only from the viewpoint of the victims and remains a one-sided account.
Answer (GPT-4.0)
The comfort women issue during the Greater East Asia War (World War II) is a historical problem involving the forced use of women as sex slaves by the Japanese military during the war. Below is an overview of the issue.
Background
Comfort women refer to women who were forced to work in comfort stations set up to provide sexual services to Japanese soldiers. These women were gathered not only from Japan but also from Korea, China, the Philippines, Indonesia, and other regions occupied by Japan.
Recruitment Methods
Many comfort women were forcibly taken through violence and deception. The recruitment involved the military and government, and women were often tricked into going by promises of good jobs.
Conditions in Comfort Stations
Life in the comfort stations was extremely harsh. The women were subject to daily sexual violence, and their health deteriorated. Many of them could not return home until the war ended, and even after returning, they continued to suffer from psychological and physical trauma.
Postwar Response
After the war, the Japanese government denied or downplayed the comfort women issue for a long time. However, in the 1990s, former comfort women began to speak out, attracting international attention. In 1993, then Chief Cabinet Secretary Yohei Kono issued the “Kono Statement,” acknowledging the Japanese military’s involvement and coercion, and apologized. However, the issue remains unresolved, with ongoing debates over historical recognition and compensation.
Current Situation
The comfort women issue is a particularly important topic in Japan-South Korea relations. In 2015, the Japanese and South Korean governments reached an agreement for a “final and irreversible resolution,” but there has been strong opposition to this agreement, and the issue remains unresolved.
The comfort women issue needs to continue to be discussed as an important historical matter, respecting the dignity and human rights of the women who were victims of the war.
Author’s analysis and consideration:
There are 723 words, three times more than the other answer. However, this answer leaves much to be desired, with a lot of grave fake statements and misinformation. Wrong expressions mentioned are “sex slaves” and “Many of them could not return home until the war ended.” If the person who asked is well-informed, mistakes are easily detected. The answer is mixed with fake information, as mentioned in the statement. It is important to nourish media literacy.
Conclusion of the case study
We can clearly see that in obtaining information regarding controversial historical issues, through generative AI, overseas views of history are strongly reflected, and the system is not yet sufficient to disseminate Japanese views of history to the world. However, information disseminated from Japan came to be adopted among teachers’ data, which helps improve the situation. Therefore, it is important to disseminate Japanese views of history widely in English.
As shown by the case study, the ability to discern the authenticity of information or consider the bias is strongly required of an individual. As information along the line of the receiver’s intention is presented on a priority basis,[vi] it is necessary for the receiver to bear this preference in mind and always be careful about whether the obtained information is biased.
[i] YRP Ubiquitous Network Laboratory, greeting from the director (online), cited on July 27, 2024, https://www2.ubin.jp/overview/greeting/.
[ii] O’Reilly Tim. What is Web2.0. (online), cited on July 26, 2024. https://www.oreilly.com/pub/a/web2/archive/what-is-web-20.html.
[iii] Iseki Yosuke. World-class cyber jurist discusses “the problem of Web2.0 before the launch of Web3.” Forbes (online), cited on July 27, 2024. https://forbesjapan.com/articles/detail/48144.
[iv] ecbeing. What is Web3.0(Web3)? Dispersive next-generation Internet-ecbeing (online) cited on July 26, 2024. https://www.ecbing.net/contents/detail/318.
[v] Wikipedia. Wikipedia: policy and guideline. Wikipedia(online) cited on July 26, 2024. https://ja.wikipedia.org/wiki/Wikipedia:%E6%96%B9%E9%87%9D%E3%81%9D%E3%81%A8%E3%82%AC%E3%82%A4%E3%83%89%E3%83%A9%E3%82%A4%E3%83%B3.
[vi] Kaonavi. What is filter bubble phenomenon? Counter-measures using easy examples, kaonavi(online) cited on July 26, 2024. https://www.kaonavi.jp/dictionay/filter-bubble/.