Flattening of Cultural Resources by Generative AI

Generative AI can learn from cultural corpora at scale and is increasingly able to evaluate, summarize, and explain creative works. This paper argues, however, that there are structural limits to using AI as a process-level partner in the conditions that produce those works. Framing this tension as the “flattening” of cultural resources, it introduces a simple Interior/Periphery model of creation and uses a representative case to illustrate a class of expressions that may be readily assessable by AI yet difficult to support during creation. The paper then notes how such flattening could feed back into future learning environments, and concludes that generative AI is best positioned at the periphery of creation rather than at its core. Afterword AI recognizes Osamu Tezuka as a great creator. Yet at the same time, it rejects the very creative process that was necessary for his works to come into being. It can evaluate the work, but it refuses to support the act of making it. This tension becomes especially visible when ethical guardrails restrict specificity about individual episodes. In doing so, it quietly reveals how difficult it is to place AI at the core of creation. AIは手塚治虫を偉大だと認める。しかし同時に、彼の作品が生まれるために必要だった制作過程を、AIは拒絶する。作品を評価できるのに制作支援を拒絶する。とりわけ倫理コードのガードレールが、個別事象への具体性を制限する局面で、そのねじれは露わになる。創作の内部にAIを置くことの難しさを、これは静かに示している。