{"stable_id":"88c7ce548e467e12","slug":"deepmind-gnome-materials-discovery","company":"Google DeepMind","workflow_name":"GNoME graph-network crystal stability discovery","function_code":"healthcare_screening","pattern_codes":["continuous_decisioning"],"changed_assumption":"Learned models can explore candidate structures beyond conventional computation and human-chosen searches.","evidence_strength":"verified","publication_tier":"showcase","freshness":"current","reviewed_at":"2026-08-23","updated_at":"2026-08-22","source_quality_summary":"1 peer reviewed; publication outcomes are verified.","caveat_summary":"Predicted thermodynamic stability is not experimental synthesis or practical utility; future discoveries can displace convex-hull entries.","freshness_caveat":null,"workflow_summary":{"before":[{"actor":"Computational materials scientist","action":"Select candidate structures and run high-cost stability calculations.","actor_type":"control"},{"actor":"Experimental scientist","action":"Choose a small subset for synthesis.","actor_type":"control"}],"hinge":"Learned models can explore candidate structures beyond conventional computation and human-chosen searches.","after":[{"actor":"GNoME graph networks","action":"Predict energies and identify candidate stable structures at large scale.","actor_type":"system"},{"actor":"Computational and experimental scientists","action":"Validate candidates with DFT, database comparison, and synthesis.","actor_type":"system"}],"decision_mode":"moved","decision_marker":"Selection moves from a fixed rule to the model"},"before":[{"order":1,"actor":"Computational materials scientist","action":"Select candidate structures and run high-cost stability calculations.","handoff_to":"Candidate database","control":"Human search strategy"},{"order":2,"actor":"Experimental scientist","action":"Choose a small subset for synthesis.","handoff_to":"Laboratory","control":"Expert judgment"}],"after":[{"order":1,"actor":"GNoME graph networks","action":"Predict energies and identify candidate stable structures at large scale.","handoff_to":"Higher-fidelity computation and databases","control":"Model hit-rate and convex-hull criteria"},{"order":2,"actor":"Computational and experimental scientists","action":"Validate candidates with DFT, database comparison, and synthesis.","handoff_to":"Materials Project and laboratories","control":"Humans retain validation and usefulness decisions"}],"decision_rights":"GNoME ranks predicted stability; scientists and higher-fidelity computation determine whether candidates are credible and worth synthesis.","exception_path":"Predictions can be displaced by future discoveries, fail higher-fidelity calculations, or prove unsynthesizable; those cases remain outside validated material knowledge.","removed_work":["Much of the brute-force screening of low-probability structures","Manual restriction of search to familiar compositions"],"outcomes":[{"metric":"New predicted stable crystal structures","baseline":"About 48,000 stable crystals in external datasets by 2023","result":"2.2 million structures stable relative to prior work; 381,000 entries on the updated convex hull","period":"Research through July 2023, published 2023-11-29","scale":"Large-scale computational exploration with about 216,000 consistent DFT calculations used for comparison","attribution_caveat":"Predicted thermodynamic stability is not experimental synthesis or practical utility; future discoveries can displace convex-hull entries.","evidence_label":"verified"}],"executive_lesson":"The workflow gain is candidate-space expansion; governance requires keeping predicted stability, successful synthesis, and useful material performance as separate gates.","anti_pattern":"Calling 381,000 predictions newly manufactured materials.","questions_for_leaders":["What fraction crosses each validation gate?","Is the bottleneck discovery, synthesis, or application testing?"],"collections":[],"bundle_version":"1.0.0","bundle_fingerprint":"sha256:c23c6cc2b88153d008ea8fda928f632ce0011fc2d4c5036672a16e5d895bab93","canonical_url":"https://brianletort.ai/transformations/deepmind-gnome-materials-discovery"}