Doubly robust inference via calibration

Name: Doubly robust inference via calibration
Uploaded: 2026-09-28T11:29:45.054405+00:00
Duration: 393 s
Description: The episode discusses the paper "Doubly robust inference via calibration," which introduces calibrated DML to fix a statistical mismatch between consistency and asymptotic normality in estimators. The hosts conclude that this method allows for doubly robust asymptotic normality, providing stronger statistical guarantees for AI inference and pushing the field toward more rigorous causal inference.

The episode discusses the paper "Doubly robust inference via calibration," which introduces calibrated DML to fix a statistical mismatch between consistency and asymptotic normality in estimators. The hosts conclude that this method allows for doubly robust asymptotic normality, providing stronger statistical guarantees for AI inference and pushing the field toward more rigorous causal inference.

Open the full episode