Causal Consistency Metrics Predict Multimodal Model Alignment Under Distributional Shift

This report synthesises findings from 9 peer-reviewed papers addressing the following research question: Can causal consistency metrics derived from Structural Causal Models serve as a reliable predictor for the alignment performance of multimodal models under distributional shift. 10 claims were extracted from source literature; 9 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 8.7/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research goal: Can causal consistency metrics derived from Structural Causal Models serve as a reliable predictor for the alignment performance of multimodal models under distributional shift?

Autonomous literature synthesis. Automated review score: 8.7/10. Full text and citation available at Assignee Research.