# Mining Causality: AI-Assisted Search for Instrumental Variables

arXiv: https://arxiv.org/abs/2409.14202 · HTML: https://mlsanity.com/paper/2409.14202/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 16 of its 69 references that are in the map.
- First posted on arXiv 21 Sep 2024.

## Builds on (most cited first, 16 of 16)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Causal Reasoning and Large Language Models: Opening a New Frontier for Causality](https://mlsanity.com/paper/2305.00050/) (2023)
- [AI Chains: Transparent and Controllable Human-AI Interaction by Chaining Large Language Model Prompts](https://mlsanity.com/paper/2110.01691/) (2021)
- [Better Zero-Shot Reasoning with Role-Play Prompting](https://mlsanity.com/paper/2308.07702/) (2023)
- [Can Large Language Models Infer Causation from Correlation?](https://mlsanity.com/paper/2306.05836/) (2023)
- [CLadder: Assessing Causal Reasoning in Language Models](https://mlsanity.com/paper/2312.04350/) (2023)
- [Can large language models build causal graphs?](https://mlsanity.com/paper/2303.05279/) (2023)
- [Machine Learning Methods Economists Should Know About](https://mlsanity.com/paper/1903.10075/) (2019)
- [Integrating Large Language Models in Causal Discovery: A Statistical Causal Approach](https://mlsanity.com/paper/2402.01454/) (2024)
- [Applied Causal Inference Powered by ML and AI](https://mlsanity.com/paper/2403.02467/) (2024)
- [Multi-Agent Causal Discovery Using Large Language Models](https://mlsanity.com/paper/2407.15073/) (2024)
- [Large Language Models for Causal Discovery: Current Landscape and Future Directions](https://mlsanity.com/paper/2402.11068/) (2024)
- [Causal Structure Learning Supervised by Large Language Model](https://mlsanity.com/paper/2311.11689/) (2023)
- [Large Language Models for Constrained-Based Causal Discovery](https://mlsanity.com/paper/2406.07378/) (2024)
- [LABOR-LLM: Language-Based Occupational Representations with Large Language Models](https://mlsanity.com/paper/2406.17972/) (2024)
- [Deep Learning for Economists](https://mlsanity.com/paper/2407.15339/) (2024)

## Built on by (most cited first, 7 of 7)

- [How Deep Do Large Language Models Internalize Scientific Literature and Citation Practices?](https://mlsanity.com/paper/2504.02767/) (2025)
- [Unveiling and Causalizing CoT: A Causal Pespective](https://mlsanity.com/paper/2502.18239/) (2025)
- [Mitigating Sycophancy in Decoder-Only Transformer Architectures: Synthetic Data Intervention](https://mlsanity.com/paper/2411.10156/) (2024)
- [Addressing the alignment problem in transportation policy making: an LLM approach](https://mlsanity.com/paper/2510.13139/) (2025)
- [IV Co-Scientist: Multi-Agent LLM Framework for Causal Instrumental Variable Discovery](https://mlsanity.com/paper/2602.07943/) (2026)
- [CausalSmith: A Formally Grounded, Self-Improving Agentic Framework for Automated Research in Causal Inference](https://mlsanity.com/paper/2607.22511/) (2026)
- [Can AI Master Econometrics? Evidence from Econometrics AI Agent on Expert-Level Tasks](https://mlsanity.com/paper/2506.00856/) (2025)

## Abstract (arXiv)

The instrumental variables (IVs) method is a leading empirical strategy for causal inference. Finding IVs is a heuristic and creative process, and justifying their validity---especially exclusion restrictions---is largely rhetorical. We propose using large language models (LLMs) to search for new IVs through narratives and counterfactual reasoning, similar to how a human researcher would. The stark difference, however, is that LLMs can dramatically accelerate this process and explore an extremely large search space. We introduce a discovery pipeline that searches for potentially novel and valid IVs using two-step and role-playing prompting strategies. We contend that these strategies simulate the endogenous decision-making of economic agents and social actors and ground language models in real-world scenarios, thereby masking the IV discovery task itself. We apply our method to three canonical areas in economics: returns to schooling, demand and supply, and peer effects. We then introduce an evaluation pipeline in which we conduct expert surveys and train an LLM judge on the survey data. This evaluation reveals that the IV candidates discovered by our method appear both novel and likely valid, whereas a more direct search tends to return instruments already established in the literature.