An art-science perspective on artificial intelligence creativity: From problem finding to materiality and embodied cognition

Abstract

Current large language models, image generators and discovery engines fuel fears that artificial intelligence systems will replace human-driven creativity. However, analysing AI systems from the perspective of creative process reveals significant limitations. Human creativity begins with finding or recognizing novel problems or challenges, which no AI system has managed. The problems AI systems address are predetermined by human users, who also provide the data and constraints bounding effective answers. Thus, human beings still carry out the vast majority of creative process-related functions for AI. Moreover, most human creativity is embodied and involves the manipulation of tools and materials. Furthermore, all human creativity is based on “tagging” information and experiences through perceptions, sensations and emotions with meanings or actions. No AI has these attributes. All human innovations also involve “untagging” preconceived meanings and actions so as to “retag” them in novel and effective ways that change how we feel, understand and act. No AI can untag or retag data, let alone act. Finally, human creative thinking is based on observing, imaging, abstracting, analogizing, playacting, modelling, body thinking, etc., of which AI systems are capable only of pattern forming and pattern recognition. Thus, the challenges for developing true AI creativity are extensive.