{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,24]],"date-time":"2026-09-24T13:08:41Z","timestamp":1790255321926,"version":"4.1.0"},"reference-count":41,"publisher":"American Chemical Society (ACS)","issue":"38","license":[{"start":{"date-parts":[[2026,9,24]],"date-time":"2026-09-24T00:00:00Z","timestamp":1790208000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,9,24]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Materials AI is advancing at an exceptional pace and attracting growing investment, while the experimental scientific facilities and people needed to test and improve its predictions face mounting pressure. Unlike increasingly standardized data-center builds, neutron facilities are bespoke, source- and instrument-specific capabilities that take many years to build and mature, making lost capacity exceptionally difficult to replace. The response should be to embrace this shift by aligning experimental capability with the problems AI is increasingly being used to solve. For atomistic materials modeling, machine-learned interatomic potentials (MLIPs) provide a natural focus, and we propose coordinated experimental validation at two scales. Workhorse campaigns would combine automated room-temperature diffraction and low-temperature inelastic neutron scattering across reusable libraries of hundreds of well-characterized samples, with observables predicted before beamtime and compared systematically with experiment. Targeted parametric studies would examine selected systems across temperature, pressure, adsorption and guest loading using quasielastic neutron scattering, total scattering and complementary diffraction and spectroscopy. Together, these measurements would expose broad performance trends and physical origins of failure, while separating development, diagnostic and blind challenge data for subsequent model improvement. Neutron facilities are not alternatives to AI investment; they are experimental infrastructure through which atomistic materials AI can be tested against reality.<\/jats:p>","DOI":"10.1021\/acs.jpclett.6c02724","type":"journal-article","created":{"date-parts":[[2026,9,24]],"date-time":"2026-09-24T12:39:48Z","timestamp":1790253588000},"page":"10775-10782","source":"Crossref","is-referenced-by-count":0,"title":["AI Can Predict,\nbut Experiments Must Test: A New Role\nfor Neutron Facilities in a Materials AI World"],"prefix":"10.1021","volume":"17","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-8326-3097","authenticated-orcid":true,"given":"Jeff","family":"Armstrong","sequence":"first","affiliation":[{"name":"Rutherford Appleton Laboratory , , Harwell Campus, ,","place":["Didcot, United Kingdom, OX11 0QX"]},{"name":"University of Bath , , ,","place":["Bath, United Kingdom, BA2 7AY"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"316","published-online":{"date-parts":[[2026,9,24]]},"reference":[{"key":"2026092408394484100_cit1","doi-asserted-by":"publisher","first-page":"10142","DOI":"10.1021\/acs.chemrev.0c01111","article-title":"Machine learning force fields","volume":"121","author":"Unke","year":"2021","journal-title":"Chem. Rev."},{"key":"2026092408394484100_cit2","doi-asserted-by":"publisher","first-page":"121501","DOI":"10.1063\/5.0139611","article-title":"How to validate machine-learned interatomic\npotentials","volume":"158","author":"Morrow","year":"2023","journal-title":"J. Chem. Phys."},{"key":"2026092408394484100_cit3","doi-asserted-by":"publisher","first-page":"104105","DOI":"10.1063\/5.0099929","article-title":"Indirect learning and physically guided validation\nof interatomic potential models","volume":"157","author":"Morrow","year":"2022","journal-title":"J. Chem. Phys."},{"key":"2026092408394484100_cit4","doi-asserted-by":"publisher","first-page":"836","DOI":"10.1038\/s42256-025-01055-1","article-title":"A Framework to Evaluate\nMachine Learning Crystal Stability\nPredictions","volume":"7","author":"Riebesell","year":"2025","journal-title":"Nature Machine Intelligence"},{"key":"2026092408394484100_cit5","doi-asserted-by":"publisher","first-page":"9","DOI":"10.1038\/s41524-024-01500-6","article-title":"Systematic softening in universal\nmachine learning interatomic potentials","volume":"11","author":"Deng","year":"2025","journal-title":"npj\nComput. Mater."},{"key":"2026092408394484100_cit6","doi-asserted-by":"publisher","first-page":"8097","DOI":"10.1021\/acs.jcim.5c01262","article-title":"MS25: materials\nscience-focused benchmark data set for machine learning interatomic\npotentials","volume":"65","author":"Maxson","year":"2025","journal-title":"J. Chem. Inf. Model."},{"key":"2026092408394484100_cit7","doi-asserted-by":"publisher","first-page":"894","DOI":"10.1038\/s43588-026-01019-4","article-title":"UniFFBench: evaluating\nuniversal machine learning force fields against experimental measurements","volume":"6","author":"Mannan","year":"2026","journal-title":"Nat. Comput. Sci."},{"key":"2026092408394484100_cit8","doi-asserted-by":"publisher","first-page":"104105","DOI":"10.1063\/5.0016004","article-title":"Committee neural network\npotentials\ncontrol generalization errors and enable active learning","volume":"153","author":"Schran","year":"2020","journal-title":"J. Chem. Phys."},{"key":"2026092408394484100_cit9","doi-asserted-by":"publisher","first-page":"20","DOI":"10.1038\/s41524-020-0283-z","article-title":"On-the-fly active learning of interpretable\nBayesian force fields for atomistic rare events","volume":"6","author":"Vandermause","year":"2020","journal-title":"npj Comput. Mater."},{"key":"2026092408394484100_cit10","doi-asserted-by":"publisher","first-page":"107206","DOI":"10.1016\/j.cpc.2020.107206","article-title":"DP-GEN: a\nconcurrent learning platform\nfor the generation of reliable deep learning based potential energy\nmodels","volume":"253","author":"Zhang","year":"2020","journal-title":"Comput. Phys. Commun."},{"key":"2026092408394484100_cit11","doi-asserted-by":"publisher","first-page":"11512","DOI":"10.1021\/acs.jpclett.2c03163","article-title":"Transferable\nforce fields from experimental\nscattering data with machine learning assisted structure refinement","volume":"13","author":"Shanks","year":"2022","journal-title":"J. Phys. Chem. Lett."},{"key":"2026092408394484100_cit12","doi-asserted-by":"publisher","first-page":"6884","DOI":"10.1038\/s41467-021-27241-4","article-title":"Learning neural network potentials\nfrom experimental\ndata via differentiable trajectory reweighting","volume":"12","author":"Thaler","year":"2021","journal-title":"Nat. Commun."},{"key":"2026092408394484100_cit13","doi-asserted-by":"publisher","first-page":"816","DOI":"10.1038\/s41467-025-56061-z","article-title":"Refining potential\nenergy surface through dynamical properties via\ndifferentiable molecular simulation","volume":"16","author":"Han","year":"2025","journal-title":"Nat. Commun."},{"key":"2026092408394484100_cit14","doi-asserted-by":"publisher","first-page":"443","DOI":"10.1016\/j.physb.2018.02.034","article-title":"AbINS: the\nmodern software for INS interpretation","volume":"551","author":"Dymkowski","year":"2018","journal-title":"Physica\nB: Condensed Matter"},{"key":"2026092408394484100_cit15","doi-asserted-by":"publisher","first-page":"030504","DOI":"10.1088\/2632-2153\/adfa68","article-title":"Benchmarking universal machine learning interatomic\npotentials for\nrapid analysis of inelastic neutron scattering data","volume":"6","author":"Han","year":"2025","journal-title":"Machine Learning: Science and Technology"},{"key":"2026092408394484100_cit16","doi-asserted-by":"publisher","first-page":"25509","DOI":"10.1039\/D5TA03325J","article-title":"Predicting neutron experiments from first principles:\na workflow powered by machine learning","volume":"13","author":"Lindgren","year":"2025","journal-title":"J. Mater.\nChem. A"},{"key":"2026092408394484100_cit17","doi-asserted-by":"publisher","first-page":"167097","DOI":"10.1016\/j.nima.2022.167097","article-title":"The unlocking of high-pressure science\nwith broadband neutron spectroscopy at the ISIS Pulsed Neutron &\nMuon Source","volume":"1039","author":"Armstrong","year":"2022","journal-title":"Nucl. Instrum. Methods Phys. Res.,\nSect. A"},{"key":"2026092408394484100_cit18","doi-asserted-by":"publisher","first-page":"10906","DOI":"10.1021\/acs.jpclett.5c01832","article-title":"Molecular Derailment via Pressurization\nin Methylammonium Lead Iodide","volume":"16","author":"Marin-Villa","year":"2025","journal-title":"J. Phys. Chem.\nLett."},{"key":"2026092408394484100_cit19","doi-asserted-by":"publisher","first-page":"184","DOI":"10.1021\/acs.jpclett.4c03336","article-title":"Methylammonium Lead Iodide across\nPhysical Space: Phase Boundaries and Structural Collapse","volume":"16","author":"Marin-Villa","year":"2025","journal-title":"J. Phys. Chem. Lett."},{"key":"2026092408394484100_cit20","doi-asserted-by":"publisher","first-page":"968","DOI":"10.1016\/j.carbon.2020.11.063","article-title":"Effect of pore geometry\non ultra-densified hydrogen in microporous carbons","volume":"173","author":"Tian","year":"2021","journal-title":"Carbon"},{"key":"2026092408394484100_cit21","doi-asserted-by":"publisher","first-page":"341","DOI":"10.1016\/0921-4526(92)90036-R","article-title":"Bayesian analysis\nof quasielastic neutron scattering data","volume":"182","author":"Sivia","year":"1992","journal-title":"Physica\nB: Condensed Matter"},{"key":"2026092408394484100_cit22","doi-asserted-by":"publisher","first-page":"4305","DOI":"10.1039\/D0CY00670J","article-title":"Molecular\nbehaviour of methanol and\ndimethyl ether in H-ZSM-5 catalysts as a function of Si\/Al ratio:\na quasielastic neutron scattering study","volume":"10","author":"Omojola","year":"2020","journal-title":"Catal.\nSci. Technol."},{"key":"2026092408394484100_cit23","doi-asserted-by":"publisher","first-page":"8411","DOI":"10.1021\/acs.jpclett.6c01399","article-title":"Lost in\nTranslation: Simulation-Informed Bayesian Inference\nImproves Characterization of Molecular Motion from Neutron Scattering","volume":"17","author":"Richardson","year":"2026","journal-title":"J. Phys. Chem. Lett."},{"key":"2026092408394484100_cit24","doi-asserted-by":"publisher","first-page":"6596","DOI":"10.1021\/acs.jpclett.6c00720","article-title":"High-pressure inelastic\nneutron spectroscopy:\nexperimental validation of machine-learned interatomic potential energy\nlandscapes","volume":"17","author":"Armstrong","year":"2026","journal-title":"J. Phys. Chem. Lett."},{"key":"2026092408394484100_cit25","doi-asserted-by":"publisher","first-page":"214325","DOI":"10.1103\/pz75-kvfy","article-title":"Collective phonon\nmixing and eigenvector transport under isotope substitution","volume":"113","author":"Armstrong","year":"2026","journal-title":"Phys. Rev. B"},{"key":"2026092408394484100_cit26","doi-asserted-by":"publisher","first-page":"658","DOI":"10.1021\/acs.jpclett.0c03292","article-title":"Vibrational\nMotions Make Significant Contributions to Sequential Methyl C\u2013H\nActivations in an Organometallic Complex","volume":"12","author":"Armstrong","year":"2021","journal-title":"J.\nPhys. Chem. Lett."},{"key":"2026092408394484100_cit27","doi-asserted-by":"publisher","first-page":"68","DOI":"10.1016\/j.nima.2018.04.009","article-title":"The neutron guide upgrade\nof the TOSCA spectrometer","volume":"896","author":"Pinna","year":"2018","journal-title":"Nucl. Instrum. Methods\nPhys. Res., Sect. A"},{"key":"2026092408394484100_cit28","doi-asserted-by":"publisher","first-page":"2938","DOI":"10.1039\/D5SU00024F","article-title":"Probing adsorption\ninteractions of lignin derivatives in industrial zeolite catalysts\nthrough combining vibrational spectroscopy and ab initio calculations","volume":"3","author":"Morton","year":"2025","journal-title":"RSC Sustainability"},{"key":"2026092408394484100_cit29","doi-asserted-by":"publisher","first-page":"012003","DOI":"10.1088\/1742-6596\/554\/1\/012003","article-title":"Recent\nand future developments on TOSCA at ISIS","volume":"554","author":"Parker","year":"2014","journal-title":"J.\nPhys.:Conf. Ser."},{"key":"2026092408394484100_cit30","doi-asserted-by":"publisher","first-page":"3756","DOI":"10.1039\/D4CY00321G","article-title":"The effect\nof molecular shape and pore structure on local and nanoscale cresol\nbehaviour in commercial zeolite catalysts","volume":"14","author":"Morton","year":"2024","journal-title":"Catal.\nSci. Technol."},{"key":"2026092408394484100_cit31","doi-asserted-by":"publisher","first-page":"11720","DOI":"10.1021\/acsanm.6c00899","article-title":"Functionalized MOFs for Subnanometric Control of Pd\nSpeciation for Selective Hydrogenation of Butadiene","volume":"9","author":"Decarolis","year":"2026","journal-title":"ACS Appl. Nano Mater."},{"key":"2026092408394484100_cit32","doi-asserted-by":"publisher","first-page":"26770","DOI":"10.1021\/acs.langmuir.5c03280","article-title":"Fine-Tuning of Gate Opening in Hybrid Zeolitic Imidazolate\nFramework-8: Inelastic Neutron Scattering and Density Functional Theory\nStudy","volume":"41","author":"Mor","year":"2025","journal-title":"Langmuir"},{"key":"2026092408394484100_cit33","doi-asserted-by":"publisher","first-page":"1290","DOI":"10.1038\/s41563-026-02578-7","article-title":"The Data-Only Illusion\nin Materials Discovery","volume":"25","author":"Smit","year":"2026","journal-title":"Nat. Mater."},{"key":"2026092408394484100_cit34","doi-asserted-by":"publisher","first-page":"301","DOI":"10.1107\/S1600576714027575","article-title":"The NeXus data format","volume":"48","author":"Ko\u0308nnecke","year":"2015","journal-title":"J. Appl. Crystallogr."},{"key":"2026092408394484100_cit35","doi-asserted-by":"publisher","first-page":"38263","DOI":"10.1021\/acsomega.6c04971","article-title":"Raw data as\nthe verification layer: why AI-driven materials discovery needs experimental\ninfrastructure","volume":"11","author":"Reeves-McLaren","year":"2026","journal-title":"ACS Omega"},{"key":"2026092408394484100_cit36","doi-asserted-by":"publisher","first-page":"17","DOI":"10.1007\/s11837-011-0102-6","article-title":"The Potential\nof Atomistic Simulations and the Knowledgebase of Interatomic Models","volume":"63","author":"Tadmor","year":"2011","journal-title":"JOM"},{"key":"2026092408394484100_cit37","doi-asserted-by":"crossref","unstructured":"Chiang, Y.; Kreiman, T.; Zhang, C.; Kuner, M. C.; Weaver, E. J.; Amin, I.; Park, H.; Lim, Y.; Kim, J.; Chrzan, D.; et al.\n          MLIP Arena: Advancing\nFairness and Transparency in Machine Learning\nInteratomic Potentials via an Open, Accessible Benchmark Platform.  Advances in Neural Information Processing Systems, Datasets\nand Benchmarks Track. 2025; NeurIPS 2025 Spotlight.","DOI":"10.52202\/085713-4410"},{"key":"2026092408394484100_cit38","doi-asserted-by":"publisher","first-page":"1083","DOI":"10.1039\/D5DD00036J","article-title":"Robotic\nintegration for end-stations at scientific user facilities","volume":"4","author":"Fernando","year":"2025","journal-title":"Digital Discovery"},{"key":"2026092408394484100_cit39","doi-asserted-by":"publisher","first-page":"31936","DOI":"10.1038\/s41598-025-17049-3","article-title":"Sample changers for\ndirect geometry neutron chopper spectrometers","volume":"15","author":"Stone","year":"2025","journal-title":"Sci. Rep."},{"key":"2026092408394484100_cit40","doi-asserted-by":"publisher","first-page":"35","DOI":"10.1021\/photonsci.5c00017","article-title":"High-Throughput Robotic GIWAXS at ALS SAXS\/WAXS\nBeamline","volume":"1","author":"Schaible","year":"2026","journal-title":"Photon Sci."},{"key":"2026092408394484100_cit41","unstructured":"Department for Science, Innovation and Technology.\n          , AI for Science\nStrategy. https:\/\/www.gov.uk\/government\/publications\/ai-for-science-strategy\/ai-for-science-strategy, 2025 (accessed 17 July 2026)."}],"container-title":["The Journal of Physical\nChemistry Letters"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/pubs.acs.org\/doi\/pdf\/10.1021\/acs.jpclett.6c02724","content-type":"application\/pdf","content-version":"vor","intended-application":"unspecified"},{"URL":"https:\/\/pubs.acs.org\/jpclcd\/article-pdf\/17\/38\/10775\/68640225\/acs.jpclett.6c02724.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/pubs.acs.org\/jpclcd\/article-pdf\/17\/38\/10775\/68640225\/acs.jpclett.6c02724.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,24]],"date-time":"2026-09-24T12:49:46Z","timestamp":1790254186000},"score":1,"resource":{"primary":{"URL":"https:\/\/pubs.acs.org\/jpclcd\/article\/17\/38\/10775\/5437757\/AI-Can-Predict-but-Experiments-Must-Test-A-New"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,24]]},"references-count":41,"journal-issue":{"issue":"38","published-online":{"date-parts":[[2026,9,24]]},"published-print":{"date-parts":[[2026,9,24]]}},"URL":"https:\/\/doi.org\/10.1021\/acs.jpclett.6c02724","relation":{},"ISSN":["1948-7185"],"issn-type":[{"value":"1948-7185","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,24]]}}}