Perspective Chapter: Toward a Quantitative Framework for Evaluating AI-Generated Creativity – A Linguistic and Computational Analysis of Language Model Outputs

As artificial intelligence (AI) language models advance rapidly, they develop new capabilities for generating human-like texts across various domains, including literature, historical narratives, and academic writing. These systems demonstrate impressive linguistic performance, yet researchers continue to debate how much of their output qualifies as “creative.” This study presents a structured analytical framework that evaluates the creativity of AI-generated texts using established linguistic and computational metrics, including lexical diversity (type-token ratio), readability (Flesch Reading Ease), and textual similarity (Jaccard similarity). The researchers used two AI language models (LLMs) to generate texts in different writing styles and analyzed them to assess linguistic richness, clarity, and originality. The findings address four key research questions regarding readability, innovation measurement, linguistic diversity, and comparative creativity in relation to human benchmarks, including the works of William Shakespeare. The study offers practical insights for organizations that integrate AI into creative tasks and contributes theoretically by advancing standardized methods for evaluating AI-generated creativity. This framework lays the groundwork for future empirical research and the development of tools that assess creativity in AI-driven language applications. Throughout this study, the term AI language models (LLMs) refers to large-scale generative text systems such as ChatGPT and Gemini.