Seeking the Unfamiliar but Memorable: Conceptual Creativity as Meta-Learning
For generative AI researchers, this work formalizes a new approach to creativity as meta-learning, but the results are limited to simple datasets (MNIST) and qualitative examples, making it incremental.
The paper introduces a meta-learning framework for conceptual creativity, where a creator generates stimuli that are unfamiliar but quickly learnable by an appraiser. Using a frozen diffusion model as creator and an autoencoder or CLIP-based appraiser, the method produces novel stylistic variations and concept compositions without additional conditioning.
What does it mean to create a new concept, rather than retrieve a familiar one? Repeatedly sampling a generative model at the same prompt produces variations with similar styles and typical content. We propose that creativity is the production of stimuli that are unfamiliar to an adaptive observer at first sight, but quickly learnable from a few exposures. We formalize this as a Creator-Appraiser pair: a Creator generates a candidate, an Appraiser adapts to it for a few inner-loop learning steps, and the Appraiser's improvement becomes the reward the Creator optimizes through. We instantiate the framework with diffusion as the Creator, an autoencoder Appraiser on MNIST, and a CLIP Appraiser with a low-rank adapter for natural images. The diffusion model remains frozen with no additional language conditioning; the meta-learning gradient is enough to produce both stylistic variations and concept compositions that the base model does not generate on its own.