H Human–AI Coevolution

Entry

How Algorithmic Confounding in Recommendation Systems Increases Homogeneity and Decreases Utility

Allison J. B. Chaney, Brandon M. Stewart, Barbara E. Engelhardt

Synopsis

Simulation studies show that data collected under algorithmic recommendations is confounded — homogenizes user behavior without increasing utility.

Keywords

·algorithmic confounding ·recommender feedback ·homogeneity ·simulation

Open paper ↗ arXiv ↗ Report issue ↗

Related entries