Computational Creativity

Computational creativity asks how computer systems can produce ideas and artifacts that are novel, useful, surprising, or otherwise valued, and how people can create with them. Approaches include language-model prompting and multi-agent search, evolutionary methods and reinforcement learning, diffusion models for images and designs, and tools that let people guide or refine generations. Researchers evaluate outputs across writing, design, games, science, and other domains using benchmarks, human judgments, and automated measures of novelty, usefulness, and diversity. Current systems can generate strong results in constrained tasks, but creativity remains hard to measure consistently, and gains in individual quality can come with reduced diversity or diminished human agency.