Artificial intelligence for geothermal energy systems: a systematic evidence map of validation, transferability and deployment readiness

Global installed geothermal power capacity is approximately 18 GWe, and geothermal plants generally provide dispatchable electricity at high capacity factors. Expansion depends on reducing the capital costs and operational risks of reservoir management, plant control and equipment maintenance, and artificial intelligence has been applied to all three. Whether that work has produced evidence sufficient for deployment has not been assessed at the level of individual studies. This review reports a PRISMA-guided systematic evidence map. Searches of Scopus and Web of Science returned 8,113 records; 977 reports entered the full-text retrieval universe, 874 were assessed at full text, and 626 peer-reviewed journal articles published between 1987 and 2026 were included, resolving to 620 independent evidence units. Each included article has a full-text-grounded extraction record produced under a rule frozen before production use, with a page or section locator for every value reported quantitatively, and readiness was derived deterministically from the coded fields rather than assigned by a reader. Two thirds of the classifiable corpus is evaluated on simulation, synthetic or laboratory data, or on field data without a meaningful held-out design. Evaluation on data external to the development set is reported in 42 of the 614 articles for which the question could be determined, and three articles report a controlled pilot; none reports shadow operation or routine deployment. Maturity is stage-specific: surface conversion contributes 179 articles and four reach field evaluation, while exploration contributes 183 and more than half do. Transferability and operational integration do not advance together, so the review reports them on separate axes. The share of articles reaching field evaluation rose from 13 per cent in 2018 to 43 per cent in 2025 while externally validated work remained in single figures each year, indicating that access to field data is spreading faster than demonstrated transfer. Systematic evidence map of 626 peer-reviewed studies on AI in geothermal energy systems, coded from full text with locators. Evaluation on data external to the development set is reported in 42 of 614 determinable articles; three report a controlled pilot and none routine deployment. Readiness is derived from coded fields by a published rule, so every assignment is reproducible from the supplementary data. Transferability and operational integration do not advance together: the pilots and the external validations sit in different use cases. Field-data access is spreading while demonstrated transfer is not, rising from 13 per cent in 2018 to 43 per cent in 2025 with externally validated work flat.