This article explores the use of systems designed to generate statistically likely outputs in creative pursuits that require divergent thinking.
LLMs (Language Models) are fundamentally designed to converge toward likely outcomes rather than diverge toward rare ones.
LLMs face challenges in divergent phases of product development, but excel in convergent phases, offering pattern recognition, consistency enforcement, and iterative improvement.
Creativity is a dance between divergence and convergence, and the application of AI tools relies on skilled human guidance to establish goals and add value.