Next Token Prediction Is a Dead End for Creativity
This addresses the problem of AI creativity limitations for researchers and practitioners in AI and creative fields, proposing a shift in paradigm.
The paper argues that next-token prediction models are fundamentally misaligned with real creativity, using battle rap as a case study to show they cannot engage in adversarial or emotionally resonant exchanges.
This paper argues that token prediction is fundamentally misaligned with real creativity. While next-token models have enabled impressive advances in language generation, their architecture favours surface-level coherence over spontaneity, originality, and improvisational risk. We use battle rap as a case study to expose the limitations of predictive systems, demonstrating that they cannot truly engage in adversarial or emotionally resonant exchanges. By reframing creativity as an interactive process rather than a predictive output, we offer a vision for AI systems that are more expressive, responsive, and aligned with human creative practice.