Causal Reasoning and Large Language Models: Opening a New Frontier for Causality
What this work does
Behaviorally probes language models on multiple causal tasks and discusses connecting text-based prior knowledge with statistical and logical causal methods.
Connection to discovery
Useful for defining which parts of causal reasoning can be elicited from pretrained knowledge and which require new data or intervention evidence.
Evidence
Inspected a dialogue with both sound and erroneous causal advice and the discussion of construct-validity threats in QA-style evaluation.
Scope and limitations
Language answers may reflect memorized associations or prompt sensitivity. A convincing argument or correct familiar edge is not new experimentally identified causality.
THE RESEARCH CONVERSATION
Discussion & reactions
Share a question, an insight, or a replication experience. Sign in with GitHub to join the conversation.