2026-12-15 2026, Volume 14 Issue 6
  • Select all
  • Innovative Perspectives
    Hongmou ZHANG

    Liyan Xu proposes that human intellectual work can be divided into interpolative and extrapolative work, holds that artificial intelligence (AI) is, at its core, an interpolation engine, and argues that interpolative work should be delegated to AI. This article proposes that extrapolation also contains another kind of activity, namely problem reframing. It is characterized by proposing new directions in the absence of existing references. Most such attempts do not hold up, while a few change the state of the relevant field. This article calls this capacity imagination and advances two arguments. First, the value of imagination arises outside existing evaluation standards, and imagination is the part of extrapolation that is hardest for AI to absorb through data accumulation and relationship fitting. Second, imagination is not a capacity that humans stably possess. After AI enters everyday work, whether the conditions for practicing it contract or are redistributed depends on the order of the workflow, and this article holds that sustaining this capacity requires corresponding social arrangements. The article discusses, in turn, the sources and costs of imagination, the contraction and redistribution of the conditions for practicing imagination under AI use, and the value judgments contained in the division between interpolation and extrapolation. It concludes that the human capacity for exploration that departs from existing standards, together with the corresponding institutional arrangements, should be protected and cultivated.