Next Token Prediction Is a Dead End for Creativity

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Olatunji, Ibukun, Sheppard, Mark
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909622456025088
author Olatunji, Ibukun
Sheppard, Mark
author_facet Olatunji, Ibukun
Sheppard, Mark
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2505_19277
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Next Token Prediction Is a Dead End for Creativity
Olatunji, Ibukun
Sheppard, Mark
Artificial Intelligence
Computation and Language
J.5; I.2.0; I.2.7
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.
title Next Token Prediction Is a Dead End for Creativity
topic Artificial Intelligence
Computation and Language
J.5; I.2.0; I.2.7
url https://arxiv.org/abs/2505.19277