Linguistic Creativity in Artificial Intelligence

The development of Large Language Models (LLMs) in Artificial Intelligence (AI) has reignited philosophical debates about the nature of creativity, particularly in linguistic contexts. LLMs generate outputs that are often syntactically fluent, semantically plausible, and contextually relevant –features that create the illusion of creativity. However, this illusion arises from statistical patterning rather than genuine intentionality or engagement in the shared social practices that ground meaning in language. This thesis critiques prevailing models of creativity –focused on cognitive processes, product-based evaluations, or audience responses- for their inadequacy in assessing linguistic creativity in AI. These models overlook the intersubjective, normative, and socially embedded dimensions central to authentic creativity. Drawing on Chomsky’s theory of linguistic competence and Wittgenstein’s conception of language as a social, rule-governed practice, the thesis redefines linguistic creativity as an activity that depends on mutual understanding, contextual responsiveness, and participation in linguistic community. It argues that linguistic creativity requires not only formal competence but also social agency, intentionality, and active engagement with others in meaningful contexts. The key contribution of this research is a philosophically grounded, non-anthropocentric framework for evaluating linguistic creativity in AI. By exposing the illusion of originality in LLM outputs and situating creativity within a social-philosophical context, the thesis provides a more robust and theoretically grounded approach to understanding the limitations of AI’s creative capacities.