Prediction-Powered Causal Inferences Advances in Neural Information Processing Systems Cadei, Riccardo Demirel, Ilker De Bartolomeis, Piersilvio Lindorfer, Lukas Cremer, Sylvia ; https://orcid.org/0000-0002-2193-3868 Schmid, Cordelia Locatello, Francesco ; https://orcid.org/0000-0002-4850-0683 In many scientific experiments, the data annotating cost constraints the pace for testing novel hypotheses. Yet, modern machine learning pipelines offer a promising solution—provided their predictions yield correct conclusions. We focus on Prediction-Powered Causal Inferences (PPCI), i.e., estimating the treatment effect in an unlabeled target experiment, relying on training data with the same outcome annotated but potentially different treatment or effect modifiers. We first show that conditional calibration guarantees valid PPCI at population level. Then, we introduce a sufficient representation constraint transferring validity across experiments, which we propose to enforce in practice in Deconfounded Empirical Risk Minimization, our new model-agnostic training objective. We validate our method on synthetic and real-world scientific data, solving impossible problem instances for Empirical Risk Minimization even with standard invariance constraints. In particular, for the first time, we achieve valid causal inference on a scientific experiment with complex recording and no human annotations, fine-tuning a foundational model on our similar annotated experiment. Neural Information Processing Systems Foundation 2025 info:eu-repo/semantics/conferenceObject doc-type:ConferenceObject ConferenceObject http://purl.org/coar/resource_type/c_5794 https://research-explorer.ista.ac.at/record/22832 Cadei R, Demirel I, De Bartolomeis P, et al. Prediction-Powered Causal Inferences. In: <i>39th Conference on Neural Information Processing Systems</i>. Vol 38. Neural Information Processing Systems Foundation; 2025:82200-82229. doi:<a href="https://doi.org/10.52202/085713-2479">10.52202/085713-2479</a> eng info:eu-repo/semantics/altIdentifier/doi/10.52202/085713-2479 info:eu-repo/semantics/altIdentifier/issn/1049-5258 info:eu-repo/semantics/altIdentifier/isbn/9798331338275 info:eu-repo/semantics/altIdentifier/arxiv/2502.06343 info:eu-repo/semantics/openAccess