In this paper we propose the use of the generative AI methods in Econometrics. Generative methods avoid the use of densities as done by MCMC. They directrix simulate large samples of observables and unobservable (parameters, latent variables) and then using high-dimensional deep learner to inform a nonlinear transport map from data to parameter inferences. Our themed apply to a wide verity or econometrics problems, including those where the latent variables are updates in deterministic fashion. Further, paper we illustrate our methodology in the field of causal inference and show how generative AI provides generalization of propensity scores. Our approach can also handle nonlinearity and heterogeneity. Finally, we conclude with the directions for future research.

翻译：本文提出将生成式人工智能方法应用于计量经济学。生成式方法避免了马尔可夫链蒙特卡洛方法中密度函数的使用，通过直接模拟大规模可观测变量与不可观测变量（参数、潜变量）的样本，并利用高维深度学习器构建从数据到参数推断的非线性传输映射。该方法适用于各类计量经济学问题，包括潜变量以确定方式更新的情形。进一步地，我们在因果推断领域阐述了该方法体系，展示了生成式人工智能如何对倾向得分进行泛化。本方法还可处理非线性和异质性。最后，我们总结了未来研究方向。