Adversarial Multiplicity: Cherry-Picked Models and the Limits of Disclosure
Predictive multiplicity -- where similarly accurate models disagree on individual predictions -- can be leveraged to optimize secondary metrics (e.g., fairness) without loss in predictive performance. However, exploiting multiplicity opens the opportunity for motivated parties (or coding agents gaming their objectives)...
Cassandra C. Chou, Benjamin Q. Huynh, Allison Koenecke et al.
· 0 citations