Evidence-Type Competition: When Can Interventional Data Teach Language Models Causal Direction?

Part of

Context Sensitivity In Generative AI