Creativity serves as a cornerstone for societal progress and innovation, but its assessment remains a complex and often subjective endeavor. With the rise of advanced generative AI models capable of tasks once reserved for human creativity, the study of AI's creative potential becomes imperative for its responsible development and application. This paper addresses the complexities in defining and evaluating creativity by introducing a new concept called Relative Creativity. Instead of trying to define creativity universally, we shift the focus to whether AI can match the creative abilities of a hypothetical human. This perspective draws inspiration from the Turing Test, expanding upon it to address the challenges and subjectivities inherent in evaluating creativity. This methodological shift facilitates a statistically quantifiable evaluation of AI's creativity, which we term Statistical Creativity. This approach allows for direct comparisons of AI's creative abilities with those of specific human groups. Building on this foundation, we discuss the application of statistical creativity in contemporary prompt-conditioned autoregressive models. In addition to defining and analyzing a measure of creativity, we introduce an actionable training guideline, effectively bridging the gap between theoretical quantification of creativity and practical model training. Through these multifaceted contributions, the paper establishes a cohesive, continuously evolving, and transformative framework for assessing and fostering statistical creativity in AI models.

翻译：创造力是社会进步与创新的基石，但其评估始终是一项复杂且常带主观性的工作。随着能够胜任曾被视为人类专属创造性任务的先进生成式AI模型兴起，研究AI的创造潜力对其负责任开发与应用至关重要。本文通过引入“相对创造力”这一新概念，揭示了在定义与评估创造力过程中面临的复杂性。我们不再试图为创造力建立普适性定义，而是将焦点转向AI能否达到假设中人类的创造力水平。这一视角借鉴了图灵测试的思路，并对其加以扩展以应对评估创造力时固有的挑战与主观性。这种方法论转变有助于实现AI创造力的统计可量化评估，我们称之为“统计创造力”。该方法能直接比较AI与特定人类群体的创造力水平。在此基础之上，我们探讨了统计创造力在当代提示条件自回归模型中的应用。除了定义并分析创造力度量标准外，本文还引入了一项可操作训练指南，有效弥合了创造力理论量化与模型实践训练之间的鸿沟。通过这些多重贡献，本文建立了一个统一、持续演进且具有变革性的框架，用于评估与培养AI模型中的统计创造力。