mlsanityData to 7 Oct 2026

InferenceEvolve: Towards Automated Causal Effect Estimators through Self-Evolving AI

arXiv 2604.04274 ↗First posted 5 Apr 2026cs.AI cs.CE stat.APcited by 0

On the map: Synthetic data and self-improvement for LLMs → Self-evolving LLM agents

Abstract (arXiv)

Causal inference is central to scientific discovery, yet choosing appropriate methods remains challenging because of the complexity of both statistical methodology and real-world data. Inspired by the success of artificial intelligence in accelerating scientific discovery, we introduce InferenceEvolve, an evolutionary framework that uses large language models to discover and iteratively refine causal methods. Across widely used benchmarks, InferenceEvolve yields estimators that consistently outperform established baselines: against 58 human submissions in a recent community competition, our best evolved estimator lay on the Pareto frontier across two evaluation metrics. We also developed robust proxy objectives for settings without semi-synthetic outcomes, with competitive results. Analysis of the evolutionary trajectories shows that agents progressively discover sophisticated strategies tailored to unrevealed data-generating mechanisms. These findings suggest that language-model-guided evolution can optimize structured scientific programs such as causal inference, even when outcomes are only partially observed.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Self-evolving LLM agents, one of 13 directions in Synthetic data and self-improvement for LLMs. Builds on 17 of its 43 references that are in the map. First posted on arXiv 5 Apr 2026.

Builds on

17 of 43 references.

References · 17

  1. 2023GPT-4 Technical Reportcited by 18,132
  2. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
  3. 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,175
  4. 2025AlphaEvolve: A coding agent for scientific and algorithmic discoverycited by 763
  5. 2017Meta-learners for Estimating Heterogeneous Treatment Effects using Machine Learningcited by 475
  6. 2015Illuminating search spaces by mapping elitescited by 406
  7. 2025The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Searchcited by 375
  8. 2016Generalized Random Forestscited by 309
  9. 2022Semiparametric doubly robust targeted double machine learning: a reviewcited by 129
  10. 2024Empowering Biomedical Discovery with AI Agentscited by 117
  11. 2017Automated versus do-it-yourself methods for causal inference: Lessons learned from a data analysis competitioncited by 106
  12. 2024Causal machine learning for predicting treatment outcomescited by 73
5 more
  1. 2023Integrating Large Language Model for Improved Causal Discoverycited by 54
  2. 2022Empirical Analysis of Model Selection for Heterogeneous Causal Effect Estimationcited by 25
  3. 2024Exploring Multi-Modal Data with Tool-Augmented LLM Agents for Precise Causal Discoverycited by 11
  4. 2025Challenges in Statistics: A Dozen Challenges in Causality and Causal Inferencecited by 9
  5. 2025ExpliCa: Evaluating Explicit Causal Reasoning in Large Language Modelscited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026CausalEvolve: Towards Open-Ended Discovery with Causal Scratchpad
  2. 2026EvoSCM: Scientific Belief Revision Through Causal Model Evolution and Experimentation
  3. 2025Causal-Copilot: An Autonomous Causal Analysis Agent
  4. 2026CausalGame: Benchmarking Causal Thinking of LLM Agents in Games
  5. 2025Causal MAS: A Survey of Large Language Model Architectures for Discovery and Effect Estimation
  6. 2026CausalArena: Benchmarking Causal Discovery in the Foundation Model Era
  7. 2026Leveraging Large Language Models for Causal Discovery: a Constraint-based, Argumentation-driven Approach
  8. 2026Why LLMs Fail at Causal Discovery and How Interventional Agents Escape