A new research paper explores the incentives for decision-makers, such as banks and employers, to offer algorithmic recourse to applicants rejected by automated systems. The paper proposes a screening model where recourse is both productive and selective, meaning it improves an applicant's value but varies in completion cost among individuals. The study suggests an optimal policy involves rejecting low-scoring applicants, offering recourse to an intermediate group, and accepting high-scoring applicants directly. AI
IMPACT This research could inform the design of fairer automated decision-making systems by clarifying the economic motivations behind offering explanations and pathways for appeal.
RANK_REASON The cluster contains an academic paper on algorithmic recourse. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →