Regression to the Mean: How LLMs May Quietly Flatten Originality Rather Than Spark an Explosion of New Ideas

This article critically examines the promise that LLMs would spark an explosion of new ideas and creativity. Instead, it argues that these models may quietly drive a regression toward the mean — flattening originality, novelty, and divergence into safe, statistically average outputs. The piece warns that we may mistake this gentle homogenization for progress, as the machine pulls all thinking toward the center rather than enabling a true Cambrian bloom of new ideas.