Discovering Crystal Structure Prediction Algorithms with an AI Co-Scientist
2026-06-22 • Machine Learning
Machine LearningArtificial Intelligence
AI summaryⓘ
The authors developed a system called HACO that helps discover new scientific algorithms by combining AI methods from different fields with human guidance. They applied this to predicting crystal structures from chemical compositions by adapting a vision-based generative model called MaskGIT. With some human-directed improvements, they created MaskGXT, which predicts crystal structures more accurately than previous methods on popular benchmarks. Their work shows that AI combined with targeted human input can help find better algorithms in fields where fast and reliable testing is possible.
crystal structure predictiongenerative modelingmasked generative modelMaskGITcrystallographic symmetryspace grouppolymorphtransformer modelMP-20 benchmarkhuman-AI collaboration
Authors
Kiyoung Seong, Nayoung Kim, Sungsoo Ahn
Abstract
We introduce Human-AI Co-discovery system (HACO) for scientific algorithm discovery through cross-domain search and sparse human steering. Starting from the goal of generating crystal structures from chemical compositions, HACO searched across generative modeling methodologies from multiple fields and identified MaskGIT, a masked generative model from vision, as a promising framework for crystal structure prediction (CSP). HACO instantiated this masked formulation as a discrete token model of crystal structure; guided by sparse high-level human objectives, it then added crystallographic symmetry tokens, space group stratified sampling for polymorph coverage, and sub-bin coordinate refinement, yielding the Masked Generative Crystal Transformer (MaskGXT). On the MP-20 polymorph split, MaskGXT reaches 79.06% match-everyone-to-reference (METRe) accuracy, compared with 70.87% for the strongest evaluated baseline. MaskGXT also attains the best match rate on standard MP-20 and MPTS-52 CSP benchmarks. These results provide evidence that, in domains offering cheap, fast, and well-aligned validation, transfer-guided interactive AI co-scientists can contribute to scientific algorithm discovery by identifying transferable modeling principles and combining them with targeted human domain guidance.