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

Researchers introduce InferenceEvolve, an AI framework using large language models to automatically discover and refine causal inference methods. The system outperformed 58 human submissions in a recent competition and demonstrates how AI can optimize complex scientific programs through evolutionary approaches.

• →InferenceEvolve uses large language models to automatically evolve causal inference estimators that outperform human-designed methods.
• →The framework's best estimator reached the Pareto frontier against 58 human submissions in a community competition.
• →The system progressively discovers sophisticated strategies tailored to specific data-generating mechanisms through evolutionary trajectories.
• →Researchers developed robust proxy objectives for real-world settings where outcomes are only partially observed.
• →The work demonstrates AI's potential to optimize structured scientific programs beyond traditional applications.