The Causal Chambers: Real Physical Systems as a Testbed for AI Methodology
Source
- Raw Markdown: paper_causal-chambers-2024.md
- PDF: paper_causal-chambers-2024.pdf
Core Claim
Causal Chambers provides real physical systems with known causal models and interventional data for AI methodology testing.
Action-Time-Series Notes
- It provides real interventional data, but not all tasks are sequential control datasets in the RL sense.
- It is valuable for evaluating whether a learned model respects interventions rather than merely forecasting passively.
- It belongs in the causal/interventional validation tier.