@phdthesis{oai:kitami-it.repo.nii.ac.jp:00009042,
 author = {NIEUWAZNY, JAGNA, STANISLAWA},
 month = {Mar},
 note = {Recent rapid developments in the field of Artificial Intelligence (AI), especially those regarding the development and implementation of artificial neural networks, surface the questions previously considered to arise much later in the future and were so far only addressed in philosophy, or its lighter derivatives, such as science fiction literature. One of such questions regards the limitations of Artificial Intelligence systems to fully grasp the importance and necessity in human lives of such subjective, spiritual and motivational phenomena as religious experience or a moral outlook.
 In this dissertation, I present the results of my research devoted to applying Artificial Intelligence to quantitative analysis of factors influencing the ethical outlook of Japanese people. I begin with presenting the background of this research. In particular, I focus on the relations between Artificial Intelligence and ethics, seen as a set of moral rules transmitted through religion, education or historical experience.
 After that, I report the results of experiments conducted in the course of my research. As an introduction to the first experiment, I provide an overview of previous cross-sectional research in the field of Religion and Technology. I then analyze how much religious vocabulary, in particular Buddhist vocabulary taken from the largest online dictionary of Buddhist terms, is present in everyday social space of Japanese people, particularly, in Japanese blog entries appearing on a popular blog service (Ameba blogs). I interpreted the level of everyday usage of Buddhist terms as appearance of such terms in the consciousness of people. I further analyzed what emotional and moral associations such contents generate. In particular, I analyzed whether expressions containing Buddhist vocabulary are considered appropriate or not from a moral point of view, as well as the emotional response of Internet users to Buddhist terminology.
 As a result of analyzing the data, I found out that Buddhist terms were in fact not absent as a theme from Japanese blogs and generated a strong emotional response. However, while the general reaction to several expressions using Buddhist terms was as expected, there were sometimes surprising twists in terms of social consequences, major discrepancies between what is perceived as ethically correct behavior between the Buddhist doctrine and the reasoning of the general population, as well as a considerate number of terms which have lost their original meaning and instead became slang expressions.
 Secondly, I focus on ethical education as a means to improve artificial companion’s conceptualization of moral decision-making process in human users. In particular, I focus on automatically determining whether changes in ethical education influenced core moral values in humans throughout the century. I analyze ethics as taught in Japan before WWII and today to verify how much the pre-WWII moral attitudes have in common with those of contemporary Japanese, to what degree what is taught as ethics in school overlaps with the general population’s understanding of ethics, as well as to verify whether a major reform of the guidelines for teaching the school subject of “ethics” at school after 1946 has changed the way common people approach core moral questions (such as those concerning the sacredness of human life). I selected textbooks used in teaching ethics at school from between 1935 and 1937, and those used in junior high schools today (2019) and analyzed what emotional and moral associations such contents generated. The analysis was performed with an automatic moral and emotional reasoning agent and based on the largest available text corpus in Japanese as well as on the resources of a Japanese digital library.
As a result, I found out that, despite changes in stereotypical view on Japan’s moral sentiments, especially due to historical events, past and contemporary Japanese share a similar moral evaluation of certain basic moral concepts, although there is a large discrepancy between how they perceive some actions to be beneficial to the society as a whole while at the same time being inconclusive when it comes to assessing the same action’s outcome on the individual performing them and in terms of emotional consequences. Some ethical categories, assessed positively before the war, while being associated with a nationalistic trend in education have also disappeared from the scope of interest of post-war society.
 Finally, in the course of a third experiment and based on the findings of the two previous ones, I try to answer a twofold question. Firstly, since the methods used for performing authorship analysis imply that an author can be recognized by the content he or she creates, I was interested in finding out whether it would be possible for an author identification solution to correctly attribute works to authors if in the course of years they have undergone a major psychological transition.
Secondly – and from the point of view of the evolution of an author’s ethical values – I checked what it would mean if the authorship attribution system encounters difficulties in detecting single authorship, hypothesizing that it could mean that historical events had had significant impact on the person’s ethical outlook. I set out to answer those questions through performing a binary authorship analysis task using a text classifier based on a pre-trained transformer model.
 As a result, I was able to confirm that in the case of texts authored by my target author, Arata Osada, in a time span of more than 10 years, while the classification accuracy drops by a large margin and is substantially lower than for texts by other non-fiction writers, the confidence of the model in its predictions remains at a similar level as in the case of a shorter time span, which means that the classifier was in many instances tricked into deciding that texts written by Arata Osada over a time span of multiple years were actually written by two different people, which in turn leads me to believe that a such a change can affect authorship analysis and historical events have great impact on a person’s ethical outlook as expressed in their writings. Based on the findings from those experiments, I outline future research goals., 近年における人工知能（AI）分野の急速な発展あ目覚ましく，ニューラルネットワークの実用化は，これまで近い将来には実現は可能と考えられてきた問題，例えば，哲学やサイエンス・フィクションでしか扱われて来なかった問題を議論の遡上に引き上た．そのひとつに，人工知能に宗教的経験や道徳的観念を把握させる問題がある．これは，日常生活における主観的、精神的動機付けとなりうる人間の倫理観を人工知能で扱おうとする問題である．本論文では，日本人の倫理観に影響を与える要因を定量的に分析するために、人工知能技術を活用することに注目した一連の研究成果を報告する．
　まず，本研究の背景を紹介し，人工知能と宗教，教育，或いは歴史的経験を通して共有される一連の道徳規範としての倫理観について分析する． 次に，研究の過程で行った実験とその結果について述べる．実験への導入として，宗教と技術の分野におけるこれまでの横断的研究の概要を説明する．そして，大規模日本語コーパス（YACISコーパス）に現れる仏教用語の出現率や仏教に基づいた倫理概念の有無，またはそれらとの感情的，道徳的関連性を分析し，現代日本人が持つ仏教教義に対する意識レベルの定量化を試みる． さらに，現存する最大規模のオンライン仏教用語事典を用いて，YACISコーパスから事典に登録された用語を含む文章を抽出，分析する．加えて，倫理推論エージェント（Moral Reasoning Agent）を抽出文に適用することで，文が持つ道徳的概念の把握と感情要素や倫理的要素の分析を行った．その結果，YACISコーパスには，仏教に基づいた倫理概念が存在していることはもちろん，現代日本人は倫理的に正しいとみなされる行動を正しい行動として意識しているが，場合によっては正しい行動を苦しいと捉えるケースもあることを明らかにした．
　以上を踏まえ，人間の持つ道徳的概念に対する倫理教育の影響を考察する．日本において，「道徳」を義務教育科目とするガイドラインの大幅改訂により，一般人の道徳の捉え方がどの程度変化したかを検証する．具体的には，①第二次世界大戦以前の「修身」と②現代日本の教育制度における「道徳」の教科書内容を比較分析し，前者の道徳概念が現代日本人の倫理観とどの程度共通しているか，または，教育において倫理的に正しいとみなされる行動が一般的な倫理の理解とどの程度共有されているかについて検証する． まず，1930年代の高校授業の目録である「修身教授録」（戦前データセット）と現代の中学授業で使用されている「私たちの道徳」（戦後データセット）から，『人生」または「命」に関連するフレーズを抽出する．次に，青空文庫収録の1930年代出版小説を対象に，それらの小説中に出現する「命」に関連するフレーズを抽出し，戦後データセット中に出現するフレーズと比較する．さらに，倫理推論エージェントを用いて，それらの内容の感情的，道徳的関連性を分析する． 分析の結果，教育制度の重大な改革にも関わらず，日本人の根本的倫理観には変化が見られず，過去においても現代においても日本人は特定の基本的道徳概念を共有していることが明らかになった． そして，日本人は，道徳的とみなされる行動が社会全体にとって有益であることを認識しながらも，それを実行する個人に対しては必ずしも楽観的結果をもたらさないことも認識していること，すなわち，「個人の利益」と「社会の利益」の相違が存在することも明らかとなった．
　最後に，日本人が持つ倫理観に影響を与える要因を吟味する．歴史的事象の影響を受けたとされる複数の著述家による著作物を詳細に分析し，道徳観の変化を定量化することを試みる．具体的には，以下の二つの問に答えるものである． 第一に，既存の著者属性認識技術を用いて，長年に渡って様々な要因（時間経過，経験，意見の変化など）によって執筆スタイルが変化した同一著者の作品を正しく認識できるかどうかを検証する． 第二に，同一著者判定が困難なケースに注目し，その原因を調査することで，歴史的事象が人間の倫理観に重大な影響を与える可能性があるという仮説を立て，これを検証する． 上記の問に答えるために，著者属性認識タスクを実行した．訓練データとして大規模日本文学コーパスである青空文庫コーパスを用い，事前学習によるトランスフォーマモデルを用いて著者属性判定器を構築した．テストデータには，太平洋戦争前後において大きく主張を変えた著名な教育史学者，長田新の著作物を用いた． 実験の結果，時間経過，経験，意見記述の変化が著者属性認識性能に影響を与えることが確認された．このことは，歴史的事象が同一著者の倫理観に大きな影響を与えることを示唆するものである．
　以上を総括すると，今後設計される人工知能システムは，倫理観を一定のものとするのではなく，道徳的推論における歴史的，宗教的，文化的、時間の経過がもたらした変化の影響を考慮したものとする必要があるという結論が得られる．},
 school = {北見工業大学},
 title = {A Study on Implementing of Culture, Religion and Time-Awareness to Machine Ethics Algorithms},
 year = {2021},
 yomi = {ニエウヴァジニ, ヤグナ, スタニスワヴァ}
}