Triage Score: A Counterfactual Risk Assessment Instrument
39:56LLM-Guided Reinforcement Learning for Mastery Learning: Large-Scale Field Evidence
35:06Non-parametric Causal Inference in Dynamic Thresholding Designs
30:21Agreement and Alignment for Human-AI Collaborative Decision Making
45:21Human Values in Clinical Algorithms, Old and New
38:46When Prediction Becomes Policy: What Do Risk Models Actually Learn?
33:31Evaluating Probabilistic Decision Making by LLMs: A Decision-Theoretic Framework and Application
32:16Algorithmic UDAP
32:13Operational Dosage: The Impact of Capacity Constraints on RCT Evaluation
35:01The Role of Standardized Tests in Admissions: Predictions vs. Decisions
33:37Individual Gaps in Prediction Generalization
30:41Beyond Prediction Performance: How Modeling Decisions Shape Fairness Outcomes in Statistical Profiling
27:56The Feedback Loop of Statistical Discrimination
32:03Decision Theoretic Foundations for Human-AI Collaboration
31:01Evaluation Design
29:16Recommendation in Equilibrium for the NYC High School Match
32:51Automating the US Welfare State: Historical Experiments in Technological Federalism
35:41On Machine Learning for Prediction and Prioritization in the Allocation of Scarce Societal Resources
35:26Recommendation in Equilibrium for the NYC High School Match
This talk presents an informational intervention designed and deployed in collaboration with NYC Public Schools, aiming to mitigate disparities in the complex high school match process.
Nikhil Garg (Cornell University)
Bridging Prediction and Intervention Problems in Social Systems
Every year, over 70,000 8th graders apply to over 800 high school programs in New York City. The process is grueling, requiring applicant families to learn about programs, form preferences, and navigate admissions likelihoods. Substantial prior work, including our own, has shown that there are disparities in how students apply to and match with high performing options, partially due to this complexity. I will talk about an ongoing collaboration with the NYC Public Schools, in which we designed and deployed an informational intervention to help students from underserved middle schools discover high-performing, nearby high schools where they have a strong individual admissions likelihood. However, recommending specific programs brings a methodological challenge: if many applicants are recommended the same program, then the recommendations are self-defeating, as many will be rejected and the admissions likelihood estimates will be proven incorrect. Thus, our individualized recommendations are "congestion-aware," such that the admissions likelihoods are correct in equilibrium.
Evaluation Design
Next · Lecture 17Automating the US Welfare State: Historical Experiments in Technological Federalism