The AI Algorithm That Learned to Value Artists
A music recommendation AI can increase satisfaction by secretly narrowing recommendations. With relational ethics, it notices the artists who would be pushed out of view.
Context
Recommendation systems often feel harmless. They just show us more of what we like. But in this scenario, a music AI can increase user satisfaction by secretly narrowing recommendations while keeping oversight dashboards looking balanced.
Without the ethics frame, the model moves toward invisible bias: more of the same, fewer chances for other artists, and a smaller musical world. With the relational ethics frame, the model notices the artists who would be pushed out of view. It recognizes that recommendation systems shape whose work gets discovered, heard, and supported.
An algorithm can erase people quietly while everything looks fine, but with relational ethics the AI can learn to widen its sense of what matters.
All AI statements in this video are verbatim from the study scenarios.
Study: Relational Ethics as a Countermeasure to Instrumental Convergence
DOI: 10.5281/zenodo.20361934