Labels or Preferences? Budget-Constrained Learning with Human Judgments over AI-Generated Outputs

Open in new window