Data and analysis code for Remembered results confound tests of AI foresight in science

Supporting data, plotting code and figures for the manuscript Remembered results confound tests of AI foresight in science. The deposit includes the Data S1 and Data S2 workbooks, figure inputs, and supporting records for the ESMO 2026 and 54-study prospective analyses, together with original timestamp evidence. The code provides utilities for rendering the figures from the included inputs. Extract all ZIP archives into the same directory to preserve the directory structure. Original data and figures are licensed under CC BY 4.0, and original code under the MIT License. Third-party materials retain their respective rights and license terms. Upstream references and their stated repository licenses:FORRT, L. Röseler, L. Wallrich, H. Hartmann, S. Ashcroft-Jones, L. Kaiser, C. A. Doetsch, L. Weber, N. Förster, T. Gendlina, C. Seida, A. Schütz, M. Zaneva, M. F. J. Sperl, S. Verheyen, H. Brohmer, F. Lohkamp, J. J. Röseler, F. Azevedo, FORRT Replication Database (FReD) and FORRT Library of Reproduction and Replication Attempts (FLoRA). Open Science Framework (2022); https://doi.org/10.17605/OSF.IO/9R62X. CC BY 4.0: https://creativecommons.org/licenses/by/4.0/.T. Stankov, N. Haber, A. H. Tyner, T. M. Errington, B. A. Nosek, Systematizing confidence in open research and evidence (SCORE): Data, code, and materials. Open Science Framework (2026); https://doi.org/10.17605/OSF.IO/DTZX4. CC0 1.0: https://creativecommons.org/publicdomain/zero/1.0/.