昨日は家系の従兄弟と次兄との３人で「わが家系史」探索の対談をしておりました。
　従兄弟は父の兄の子で、わたしと次兄は父の子。家系史を探るのにも、こういう「立場」の違いによる認識の相違が発生してくる。そしてそれぞれ独自に家系探査してきているので、こうした情報交換によって、格差・相違を埋め合わせたいと思ったのです。
　事実、わたしよりも12歳年長の従兄弟はその「母系」が醸成する地域社会からの情報が非常に大きい興味領域を占めていることが話していて伝わってきた。わたしたち兄弟はそういう社会とは距離を持って生きた父の生き方に沿って見てきていたので、それは非常に新鮮だった。要するに父を超えて祖父の生き様の実相、異なる見方が伝わってきた次第。父の代の兄弟でもものごとのとらえ方、祖父の事跡への解釈の仕方が「枝分かれ」していることが、実感された次第。
　そういうことなので、興味を持っている地域性・地域社会の擦り合わせがアタマのなかで高速で「仕分け・理解」しながらの対談内容。正直、従兄弟の発する地名・人名の名詞自体に対しての仕分けがまったく追いつかないのですね。
　また同時に当方からの探究の疑問点は、従兄弟も「初めて聞いた」みたいな反応が大きかったりした。・・・そういうことでなかなかに「スリリング」な展開（笑）。
　で、親戚らしく従兄弟の家に着いたらすぐに対談が始まったので、iPhoneの「ボイスメモ」を起動させたのも開始から30分程度経過後。そこから４時間超の対談記録を録音。
　さすがに昨日は疲れたので帰宅後バタンキュー。本日早朝に「整理整頓」作業に取り組んで、ボイスメモからの「文字起こし」を行った。iOSの進化ぶりはAI活用を受けて隔世の感。４時間超の対談記録があっという間に65,000字程度にテキスト化できた。
　そこからの対談内容の整理整頓（だれがしゃべった内容かの仕分けと地名・人名の「系統付け」など）もAIを活用することで、道筋がハッキリしてきた。
　いまはAIに整理作業させているところですが、AIからはやや「忖度」気味に「地域史・家族史の一次資料として十分」というような返答が帰ってきている。こういう対談についての、AI活用のひとつのテストケースにもと考えています。
　AIによる架空の事実を正しいかのように自信満々に出力してしまう現象〜ハルシネーションに気を付けて進行させたいと思います。さて・・・。

＜写真は熊本県の国宝「通潤橋」の放水シーン＞

●お知らせ
拙書「作家と住空間」幻冬舎から電子書籍で発刊
お求めはAmazonで。
https://amzn.asia/d/eUiv9yO

English version⬇

[Compiling a 4-Hour-Plus Family History Discussion on the Latest iOS and Mac into a Document]
Even among close cousins, there are differences in perspective regarding family history. Aligning our respective areas of interest. Promoting mutual understanding and integrating our perspectives. …

　Yesterday, the three of us—my cousin on my father’s side, my second oldest brother, and I—had a conversation exploring “our family history.”
My cousin is the son of my father’s older brother, while my second oldest brother and I are my father’s sons. Even when researching family history, differences in perspective arise due to these differing “positions.” Since each of us has been conducting our own family history research, I wanted to bridge the gaps and differences through this exchange of information.
　In fact, as we talked, it became clear that my cousin—who is 12 years older than me—is deeply interested in information derived from the local community shaped by his “maternal lineage.” Since my brother and I had viewed things through the lens of our father’s lifestyle—which kept him at a distance from that kind of community—this perspective was very refreshing. In short, I gained a different perspective on the true nature of my grandfather’s life, one that went beyond my father’s view. I truly came to realize that even among my father’s siblings, their ways of perceiving things and interpreting my grandfather’s deeds had “branched off” in different directions.
　Because of this, our conversation unfolded as I rapidly “sorted and understood” the regional characteristics and local communities I was interested in, all while the discussion was taking place. To be honest, I simply couldn’t keep up with sorting through all the place names and people’s names my cousin was mentioning.
　At the same time, when I asked probing questions, my cousin often reacted as if he were hearing the answers for the first time. …So, it turned out to be quite a “thrilling” exchange (laughs).
　As is typical among relatives, the conversation started the moment we arrived at my cousin’s house, so I didn’t launch the “Voice Memos” app on my iPhone until about 30 minutes into the conversation. From there, I recorded over four hours of our conversation.
　I was understandably exhausted yesterday, so I crashed as soon as I got home. Early this morning, I got to work “organizing” the recording and transcribed the Voice Memos. Thanks to AI, iOS has evolved so much that it feels like a world of difference. The over four-hour conversation was transcribed into about 65,000 characters in the blink of an eye.
　By using AI to organize the content of the conversation—such as sorting who said what and categorizing place names and people’s names—the structure has become much clearer.
I’m currently having the AI handle the organization, and it’s responding with a slightly “overly accommodating” tone, suggesting that the recording is “sufficient as primary source material for local and family history.” I view this as a test case for using AI with this type of conversation.
　I want to proceed while being careful to avoid the phenomenon where AI confidently outputs fictional facts as if they were true—a phenomenon known as “hallucination.” Well then…

● Announcement
My book, “Writers and Living Spaces,” has been published as an e-book by Gentosha.
Available on Amazon.

Posted on 7月 16th, 2026 by 三木 奎吾
Filed under: 未分類