Sep 2026· Zenodo (CERN European Organization for Nuclear Research)
Geological Modeling and Analysis
Abstract
AbyssGeist (探渊索隐) is a conceptual Geoscience Foundation Model project proposal developed around the integration of artificial intelligence, geophysical and geochemical exploration, geological knowledge, and mineral resource development. The project proposes a three-layer architecture consisting of a Multimodal Geophysical Encoder, a Geological Prior-Knowledge Graph, and a Twin Inversion & Reasoning Engine. Its core strategy is to integrate heterogeneous geoscience data, including seismic data, geophysical maps, core images, and geochemical information, into a unified representation space; retrieve historically analogous geological and geophysical data; and use geological and physical priors to constrain the inherently non-unique inversion problem. Beyond geophysical interpretation, AbyssGeist envisions a full-lifecycle digital twin connecting exploration, drilling, mining, processing, transportation, cost control, and mineral pricing. Real-time drilling and Logging While Drilling (LWD) data are envisioned as feedback signals for continuously updating and refining exploration models. The proposal therefore explores the possibility of extending a geoscience AI system from subsurface interpretation toward integrated decision-making across the mineral resource value chain. The project also includes a cultural component, “Questions to Heaven · The Voice of the Earth,” which reinterprets selected passages from Qu Yuan’s Tianwen through the lens of modern geophysical exploration. It presents a dialogue between ancient questions about the unknown Earth and contemporary technologies used to observe, model, and interpret the subsurface. This record documents the conceptual framework, technical architecture, workflow, and cultural vision of AbyssGeist. It is a project proposal and conceptual design, rather than a report of a completed foundation model or operational system.
GAOKAO-Bench is introduced, an intuitive benchmark that employs questions from the Chinese GAOKAO examination as test samples, including both subjective and objective questions that contribute a robust evaluation benchmark for future large language models and offers valuable insights into the advantages and limitations of such models.
Xiaotian Zhang, Chun-yan Li, Yi Zong et al.· arXiv.org· 216 citations· ⚡17
This work investigates the possibilities of using LLMs in a resume screening setting via a document retrieval framework that simulates job candidate selection and finds that the MTEs are biased, significantly favoring White-associated names in 85% of cases and female-associated names in only 11.1% of cases.
Empirically, PRISM reduces the end-to-end time for data selection and model tuning to just 30% of conventional pipelines, and achieves this efficiency while simultaneously enhancing performance, surpassing models fine-tuned on the full dataset across eight multimodal and three language understanding benchmarks.
Jinhe Bi, Yifan Wang, Danqi Yan et al.· arXiv.org· 73 citations· ⚡4
The method, ECCOLA, is presented, which aims at making the high-level AI ethics principles more practical, making it possible for developers to more easily implement them in practice.
Ville Vakkuri, Kai-Kristian Kemell, P. Abrahamsson· EUROMICRO Conference on Soft...· 64 citations· ⚡6
This paper designs Markov decision processes (MDPs) for different combinatorial problems and proposes to train conditional GFlowNets to sample from the solution space and demonstrates that GFlowNet policies can efficiently find high-quality solutions.
Dinghuai Zhang, H. Dai, Esmeralda S. Whitammer et al.· Advances in Neural Informati...· 59 citations· ⚡8
An empirical study on the current state of practice in artificial intelligence ethics is conducted by means of a multiple case study of five case companies, which indicates a gap between research and practice in the area.
Ville Vakkuri, Kai-Kristian Kemell, Joni Kultanen et al.· arXiv.org· 56 citations· ⚡6