Japanese scholars warn of 'knowledge appropriation' as AI trains on rare books
Debate over digitizing rare manuscripts for AI training risks losing access to original sources and erasing unique texts.
What to know
- Japanese scholars argue that training AI on rare and unique books risks erasing irreplaceable historical sources if physical access is lost.
- Academic researchers emphasize that scholarly verification requires consulting original primary sources, which AI training cannot replace.
- The debate highlights tension between making knowledge available to AI systems and preserving physical access to rare manuscripts for future scholarship.
The dispute Whether the benefits of making rare books available to AI systems outweigh the risks of losing permanent physical access to unique or out-of-print sources. · positions read across 2 posts and comments
AI training on rare books without preserving original access amounts to knowledge appropriation and risks losing unique historical sources forever.
-
“No matter how much old literature is fed to AI, if you can no longer access the original paper books, that is nothing but knowledge appropriation.”
@kotonoha_yakata · X ↗
Scholarly work requires original sources for verification, and AI training cannot substitute for access to primary texts.
-
“This is exactly right—especially for history papers, you must consult the original to ensure reliability. Feeding it to AI doesn't help.”
Hiroshi Hiranakaisseis1 · X ↗
@kotonoha_yakata Japanese commentatorHiroshi Hiranakaisseis1 Academic researcher
How it unfolded 2 developments, newest first · click a bar or a number to jump posts
-
2
Academic researchers echo concerns about scholarly verification
Scholars respond that historical research requires access to original sources to verify credibility, and that feeding rare books to AI does not substitute for this requirement. They note that losing rare copies removes the ability to verify information at all.
“This is exactly right—especially for history papers, you must consult the original to ensure reliability. Feeding it to AI doesn't help, and if rare books are consumed, there's no way to verify anything.”
— Hiroshi Hiranakaisseis1 (academic commenter) -
K
八神夕歌(やがみ ゆうか)［T.Ishiyama］さん: 「どれだけ古書籍をAIに学習させたとしても、もとの紙の書籍に当たれなくなるのであれば、それは知識の簒奪に他ならない、というのが私の立場です。その中で、もしも孤本を自炊されてしまったら、どうするつもりなのでしょうか。それは恐ろしいことです。」 / X https:// x.com/kotonoha_yakata/status/2 102592108171543022 広中一成さん: 「これはおっしゃるとおりで、(特に歴史系)論文を書くにしても信頼性を担保するには原本に当たらなければならず、AIに食わせたところでしょうがないし、稀覯本食われたら確認のしようがないわね。」 / X https://…
-
-
1
Scholar frames AI training on rare books as 'knowledge appropriation'
A Japanese commentator argues that training AI on old books constitutes appropriation of knowledge if the original physical texts become inaccessible, raising specific concern about rare unique copies being digitized for training without preserving access to originals.
“No matter how much old literature is fed to AI, if you can no longer access the original paper books, that is nothing but knowledge appropriation. If unique copies get self-scanned, what then?”
— @kotonoha_yakata -
どれだけ古書籍をAIに学習させたとしても、もとの紙の書籍に当たれなくなるのであれば、それは知識の簒奪に他ならない、というのが私の立場です。その中で、もしも孤本を自炊されてしまったら、どうするつもりなのでしょうか。それは恐ろしいことです。
-