Evaluating Interventional Reasoning Capabilities of Large Language Models
This addresses the need to evaluate LLMs for automated decision-making tasks that require causal reasoning, though it is incremental as it builds on existing work on causal fact retrieval.
The paper tackled the problem of assessing whether large language models (LLMs) can reason about causal interventions, by creating benchmarks across diverse causal graphs and evaluating six LLMs, finding that GPT models showed promising accuracy in predicting intervention effects.
Numerous decision-making tasks require estimating causal effects under interventions on different parts of a system. As practitioners consider using large language models (LLMs) to automate decisions, studying their causal reasoning capabilities becomes crucial. A recent line of work evaluates LLMs ability to retrieve commonsense causal facts, but these evaluations do not sufficiently assess how LLMs reason about interventions. Motivated by the role that interventions play in causal inference, in this paper, we conduct empirical analyses to evaluate whether LLMs can accurately update their knowledge of a data-generating process in response to an intervention. We create benchmarks that span diverse causal graphs (e.g., confounding, mediation) and variable types, and enable a study of intervention-based reasoning. These benchmarks allow us to isolate the ability of LLMs to accurately predict changes resulting from their ability to memorize facts or find other shortcuts. We evaluate six LLMs on the benchmarks, finding that GPT models show promising accuracy at predicting the intervention effects.