Current status and development trends of generative AI technology in Earth science research

doi:10.13745/j.esf.sf.2025.4.68

Abstract

Abstract:

The multidimensional development of Earth’s complex giant system research continues to drive the breakthrough innovation of geoscientific research methods. As an emerging research tool, generative AI technology provides new ideas for geoscientific research by virtue of its powerful data processing and knowledge reasoning capabilities. In this paper, we systematically sort out the development of generative AI technology, compare the technical routes of ChatGPT and the domestic DeepSeek model, and show the advantages of generative AI technology’s powerful natural language processing capability and multimodal model construction. This paper summarizes in detail the application of generative AI technology in geosciences, especially the current status and development trend of its application in the field of resource exploration. Generative AI technology has realized the universal integration of heterogeneous data from multiple sources in the field of geosciences, and has been deeply involved in various stages of work in the field of resource exploration and provided technical support at the levels of data integration, cognitive reasoning, application services, etc., which has set off a wave of innovation in the field of resource exploration. At present, generative AI technology in the field of geoscience applications still exists in data completeness defects, geological complex system challenges, geological model interpretability problems and other core constraints. Comprehensive analysis points out that generative AI reconstructs the technical system of “data perception-knowledge refinement-decision generation”, which will certainly accelerate the realization of breakthroughs in the application of this technology in the fields of Earth science, and provide important technical support for the innovation of exploration technology and methodology, improvement of exploration efficiency, and assisting the new round of strategic action of finding mineral breakthroughs to safeguard the country’s energy security.

Key words: artificial intelligence, generative AI technology, ChatGPT, ore prospecting, resource exploration, DeepSeek

CLC Number:

P628.1

ZHOU Shengquan, LI Yike, WANG Yongzhi, LIU Haiming, LI Nan, KE Changhui, LI Ruiping, ZHAO Yonggang, ZHANG Li. Current status and development trends of generative AI technology in Earth science research[J]. Earth Science Frontiers, 2025, 32(4): 303-316.

Figures/Tables 8

References 107

[1]	朱日祥, 侯增谦, 郭正堂, 等. 宜居地球的过去、现在与未来: 地球科学发展战略概要[J]. 科学通报, 2021, 66(35): 4485-4490.
[2]	丁仲礼. 固体地球科学研究方法[M]. 北京: 科学出版社, 2013.
[3]	钱学森. 再谈开放的复杂巨系统[J]. 模式识别与人工智能, 1991, 4(1): 1-4.
[4]	郝斌飞, 韩旭军, 马明国, 等. Google Earth Engine在地球科学与环境科学中的应用研究进展[J]. 遥感技术与应用, 2018, 33(4): 600-611. DOI
[5]	李灿锋, 刘达, 周德坤, 等. 人工智能在地质领域的应用与展望[J]. 矿物岩石地球化学通报, 2022, 41(3): 668-677.
[6]	蔡昌, 王艺琳. ChatGPT应用热潮的冷思考: 局限性、风险与价值挑战[J]. 商业会计, 2023(14): 12-18.
[7]	王耀祖, 李擎, 戴张杰, 等. 大语言模型研究现状与趋势[J]. 工程科学学报, 2024, 46(8): 1411-1425.
[8]	ZHOU W, ZHU X G, HAN Q L, et al. The security of using large language models: a survey with emphasis on ChatGPT[J]. IEEE/CAA Journal of Automatica Sinica, 2025, 12(1): 1-26.
[9]	尹宝才, 王文通, 王立春. 深度学习研究综述[J]. 北京工业大学学报, 2015, 41(1): 48-59.
[10]	王正龙, 张保稳. 生成对抗网络研究综述[J]. 网络与信息安全学报, 2021, 7(4): 68-85.
[11]	GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial networks[J]. arXiv e-prints, 2014. DOI: arXiv:1406.2661.
[12]	RADFORD A, NARASIMHAN K, SALIMANS T, et al. Improving language understanding by generative pre-training[EB/OL]. [2024-06-21]. https://www.mikecaptain.com/resources/pdf/GPT-1.pdf.
[13]	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[J]. arXiv e-prints, 2017. DOI: arXiv:1706.03762.
[14]	TAY Y, DEHGHANI M, BAHRI D, et al. Efficient transformers: a survey[J]. ACM Computing Surveys, 2023, 55(6): 1-28.
[15]	SHAZEER N, MIRHOSEINI A, MAZIARZ K, et al. Outrageously large neural networks: the sparsely-gated mixture-of-experts layer[EB/OL]. (2017-01-23)[2025-02-25]. https://arxiv.org/abs/1701.06538v1.
[16]	LI L J, DONG B W, WANG R H, et al. SALAD-bench: a hierarchical and comprehensive safety benchmark for large language models[EB/OL]. (2024-06-07)[2025-02-25]. https://arxiv.org/abs/2402.05044v4.
[17]	武俊宏, 赵阳, 宗成庆. ChatGPT 能力分析与未来展望[J]. 中国科学基金, 2023, 37(5): 735-742.
[18]	OUYANG L, WU J, JIANG X, et al. Training language models to follow instructions with human feedback[J]. Advances in Neural Information Processing Systems, 2022, 35: 27730-27744.
[19]	郭亚军, 李天祥, 王会森, 等. 从结绳记事到认知引擎: 人类知识增强的历史演进与DeepSeek的创新启示[J/OL]. 图书馆论坛, 2025: 1-11(2025-02-25)[2025-04-23]. https://kns.cnki.net/kcms/detail/44.1306.G2.20250224.1742.007.html.
[20]	DeepSeek-AI, GUO D Y, YANG D J, et al. DeepSeek-R1: incentivizing reasoning capability in LLMs via reinforcement learning[EB/OL]. (2025-01-22)[2025-02-25]. https://arxiv.org/abs/2501.12948v1.
[21]	BROWN T, MANN B, RYDER N, et al. Language models are few-shot learners[J]. Advances in Neural Information Processing Systems, 2020, 33: 1877-1901.
[22]	秦涛, 杜尚恒, 常元元, 等. ChatGPT的工作原理、关键技术及未来发展趋势[J]. 西安交通大学学报, 2024, 58(1): 1-12.
[23]	吴宣志. 专家系统与智能系统[J]. 物探化探计算技术, 1997, 19(2): 108-114.
[24]	张健挺, 邱友良. 人工智能和专家系统在地学中的应用综述[J]. 地理科学进展, 1998, 17(1): 44-51.
[25]	邓志东. 关于发展我国人工智能技术与产业的建议[J]. 科技导报, 2016, 34(7): 12-13.
[26]	AZARIA A, AZOULAY R, RECHES S. ChatGPT is a remarkable tool: for experts[J]. Data Intelligence, 2024, 6(1): 240-296.
[27]	李艳, 金皓月, 杨玉辉. 基于 ChatGPT 的研究生人机协同学术写作实践研究及启示[J]. 远程教育杂志, 2023, 41(5): 38-48, 75.
[28]	NAZIR A, WANG Z. A comprehensive survey of ChatGPT: advancements, applications, prospects, and challenges[J]. Meta-Radiology, 2023, 1(2): 100022.
[29]	ALI KHOWAJA S, KHUWAJA P, DEV K, et al. ChatGPT needs SPADE (sustainability, PrivAcy, digital divide, and ethics) evaluation: a review[J]. Cognitive Computation, 2024, 16(5): 2528-2550.
[30]	FOROUMANDI E, MORADKHANI H, SANCHEZ-VILA X, et al. ChatGPT in hydrology and earth sciences: opportunities, prospects, and concerns[J]. Water Resources Research, 2023, 59(10): e2023WR036288.
[31]	WILSON M P, FOULGER G R, WILKINSON M W, et al. Artificial intelligence and human-induced seismicity: initial observations of ChatGPT[J]. Seismological Research Letters, 2023, 94(5): 2111-2118.
[32]	汤井田, 任政勇, 化希瑞. 地球物理学中的电磁场正演与反演[J]. 地球物理学进展, 2007, 22(4): 1181-1194.
[33]	DOZONO K, GASIBA T E, STOCCO A. Large language models for secure code assessment: a multi-language empirical study[EB/OL]. (2024-08-24)[2025-02-25]. https://arxiv.org/abs/2408.06428v1.
[34]	VAGHEFI S A, STAMMBACH D, MUCCIONE V, et al. ChatClimate: grounding conversational AI in climate science[J]. Communications Earth & Environment, 2023, 4: 480.
[35]	匡立春, 刘合, 任义丽, 等. 人工智能在石油勘探开发领域的应用现状与发展趋势[J]. 石油勘探与开发, 2021, 48(1): 1-11. DOI
[36]	宋辉, 高洋, 陈伟, 等. 基于卷积降噪自编码器的地震数据去噪[J]. 石油地球物理勘探, 2020, 55(6): 1210-1219, 1160-1161.
[37]	于登云, 张哲, 泮斌峰, 等. 深空探测人工智能技术研究与展望[J]. 深空探测学报, 2020, 7(1): 11-23.
[38]	李硕, 吴园涛, 李琛, 等. 水下机器人应用及展望[J]. 中国科学院院刊, 2022, 37(7): 910-920.
[39]	刘双, 胡祥云, 郭宁, 等. 无人机航磁测量技术综述[J]. 武汉大学学报(信息科学版), 2023, 48(6): 823-840.
[40]	底青云, 朱日祥, 薛国强, 等. 我国深地资源电磁探测新技术研究进展[J]. 地球物理学报, 2019, 62(6): 2128-2138. DOI
[41]	SHABDIROVA A, KOZHAGULOVA A, MINH N H, et al. Application of machine learning to predict transient sand production in the Karazhanbas oil field, Ustyurt-Buzachi Basin (West Kazakhstan)[J]. Natural Resources Research, 2023, 32(5): 1975-1986.
[42]	ACOSTA I C C, KHODADADZADEH M, TOLOSANA-DELGADO R, et al. Drill-core hyperspectral and geochemical data integration in a superpixel-based machine learning framework[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 13: 4214-4228.
[43]	WANG P, SU S G, WANG G Z, et al. Discrimination of deposit types using magnetite geochemistry based on machine learning[J]. Ore Geology Reviews, 2024, 170: 106107.
[44]	BAIDYA A S, MAITI G, MONDAL S, et al. Biotite chemistry as an indicator of hydrothermal deposit types and fluid sources: insights from big data compilation, multivariate statistical analysis, and machine learning[J]. Journal of Geochemical Exploration, 2024, 259: 107442.
[45]	ZHU X F, ZHANG C, HUANG X W, et al. Principal component analysis of mineral and element composition of ores from the Bayan Obo Nb-Fe-REE deposit: implication for mineralization process and ore classification[J]. Ore Geology Reviews, 2024, 167: 105972.
[46]	PETRELLI M, PERUGINI D. Solving petrological problems through machine learning: the study case of tectonic discrimination using geochemical and isotopic data[J]. Contributions to Mineralogy and Petrology, 2016, 171(10): 81.
[47]	UEKI K, HINO H, KUWATANI T. Geochemical discrimination and characteristics of magmatic tectonic settings: a machine-learning-based approach[J]. Geochemistry, Geophysics, Geosystems, 2018, 19(4): 1327-1347.
[48]	ZHONG S H, LI S Z, LIU Y, et al. I-type and S-type granites in the Earth’s earliest continental crust[J]. Communications Earth & Environment, 2023, 4: 61.
[49]	HASTEROK D, GARD M, BISHOP C M B, et al. Chemical identification of metamorphic protoliths using machine learning methods[J]. Computers & Geosciences, 2019, 132: 56-68.
[50]	LIU H M, HARRIS J, SHERLOCK R, et al. Mineral prospectivity mapping using machine learning techniques for gold exploration in the Larder Lake area, Ontario, Canada[J]. Journal of Geochemical Exploration, 2023, 253: 107279.
[51]	MAEPA F M, SMITH R S. Examining the controls on gold deposit distribution in the Swayze greenstone belt, Ontario, Canada, using multi-scale methods of spatial data analysis[J]. Ore Geology Reviews, 2020, 125: 103671.
[52]	QIU K F, ZHOU T, CHEW D, et al. Apatite trace element composition as an indicator of ore deposit types: a machine learning approach[J]. American Mineralogist, 2024, 109(2): 303-314.
[53]	ZHENG Y Y, XU B, LENTZ D R, et al. Machine learning applied to apatite compositions for determining mineralization potential[J]. American Mineralogist, 2024, 109(8): 1394-1405.
[54]	SAHA R, UPADHYAY D, MISHRA B. Discriminating tectonic setting of igneous rocks using biotite major element chemistry: a machine learning approach[J]. Geochemistry, Geophysics, Geosystems, 2021, 22(11): e2021GC010053.
[55]	WANG Y, QIU K F, MÜLLER A, et al. Machine learning prediction of quartz forming-environments[J]. Journal of Geophysical Research: Solid Earth, 2021, 126(8): e2021JB021925.
[56]	NATHWANI C L, WILKINSON J J, BROWNSCOMBE W, et al. Mineral texture classification using deep convolutional neural networks: an application to zircons from porphyry copper deposits[J]. Journal of Geophysical Research: Solid Earth, 2023, 128(2): e2022JB025933.
[57]	ZHONG S H, LIU Y, LI S Z, et al. A machine learning method for distinguishing detrital zircon provenance[J]. Contributions to Mineralogy and Petrology, 2023, 178(6): 35.
[58]	FILZMOSER P, HRON K. Robust methods for compositional data[C]// Proceedings of COMPSTAT’2010. Heidelberg: Physica-Verlag HD, 2010: 79-88.
[59]	FILZMOSER P, HRON K, REIMANN C. Principal component analysis for compositional data with outliers[J]. Environmetrics, 2009, 20(6): 621-632.
[60]	GRUNSKY E C, DREW L J, WOODRUFF L G, et al. Statistical variability of the geochemistry and mineralogy of soils in the Maritime Provinces of Canada and part of the Northeast United States[J]. Geochemistry: Exploration, Environment, Analysis, 2013, 13(4): 249-266.
[61]	HRON K, TEMPL M, FILZMOSER P. Exploratory compositional data analysis using the R-package robCompositions[C]// Proceedings of the ninth international conference on Data Analysis and Modeling. Minsk, Belarus: Russian Academy of Sciences, 2010: 179-186.
[62]	HRON K, TEMPL M, FILZMOSER P. Imputation of missing values for compositional data using classical and robust methods[J]. Computational Statistics & Data Analysis, 2010, 54(12): 3095-3107.
[63]	ÇETIN Ş B. Real-ESRGAN: a deep learning approach for general image restoration and its application to aerial images[J]. Advanced Remote Sensing, 2023, 3(2): 90-99.
[64]	SMITH L, HORROCKS T, HOLDEN E J, et al. Magnetic grid resolution enhancement using machine learning: a case study from the Eastern Goldfields Superterrane[J]. Ore Geology Reviews, 2022, 150: 105119.
[65]	CHAPMAN R, MORTENSEN J K, MURPHY R. Compositional signatures of gold from different deposit types in British Columbia, Canada[J]. Minerals, 2023, 13(8): 1072.
[66]	CHAPMAN R, TORVELA T, SAVASTANO L. Insights into regional metallogeny from detailed compositional studies of alluvial gold: an example from the loch Tay area, central Scotland[J]. Minerals, 2023, 13(2): 140.
[67]	CHAPMAN R J, MOLES N R, BLUEMEL B, et al. Detrital gold as an indicator mineral[J]. Geological Society, London, Special Publications, 2022, 516(1): 313-336.
[68]	LIU H M, BEAUDOIN G. Geochemical signatures in native gold derived from Au-bearing ore deposits[J]. Ore Geology Reviews, 2021, 132: 104066.
[69]	LIU H M, BEAUDOIN G, MAKVANDI S, et al. Multivariate statistical analysis of trace element compositions of native gold from orogenic gold deposits: implication for mineral exploration[J]. Ore Geology Reviews, 2021, 131: 104061.
[70]	CAO G S, ZHANG Y, ZHAO H T, et al. Trace element variations of pyrite in orogenic gold deposits: constraints from big data and machine learning[J]. Ore Geology Reviews, 2023, 157: 105447.
[71]	GREGORY D D, CRACKNELL M J, LARGE R R, et al. Distinguishing ore deposit type and barren sedimentary pyrite using laser ablation-inductively coupled plasma-mass spectrometry trace element data and statistical analysis of large data sets[J]. Economic Geology, 2019, 114(4): 771-786.
[72]	ZHONG R C, DENG Y, LI W B, et al. Revealing the multi-stage ore-forming history of a mineral deposit using pyrite geochemistry and machine learning-based data interpretation[J]. Ore Geology Reviews, 2021, 133: 104079.
[73]	CARABALLO E, BEAUDOIN G, DARE S, et al. Trace element composition of chalcopyrite from volcanogenic massive sulfide deposits: variation and implications for provenance recognition[J]. Economic Geology, 2023, 118(8): 1923-1958.
[74]	CARABALLO E, DARE S, BEAUDOIN G. Variation of trace elements in chalcopyrite from worldwide Ni-Cu sulfide and Reef-type PGE deposits: implications for mineral exploration[J]. Mineralium Deposita, 2022, 57(8): 1293-1321.
[75]	LI X M, ZHANG Y X, LI Z K, et al. Discrimination of Pb-Zn deposit types using sphalerite geochemistry: new insights from machine learning algorithm[J]. Geoscience Frontiers, 2023, 14(4): 101580.
[76]	ZHAO H T, SHAO Y J, ZHANG Y, et al. Big data mining on trace element geochemistry of sphalerite[J]. Journal of Geochemical Exploration, 2023, 252: 107254.
[77]	SCIUBA M, BEAUDOIN G, MAKVANDI S. Chemical composition of tourmaline in orogenic gold deposits[J]. Mineralium Deposita, 2021, 56(3): 537-560.
[78]	MIRANDA A C R, BEAUDOIN G, ROTTIER B. Scheelite chemistry from skarn systems: implications for ore-forming processes and mineral exploration[J]. Mineralium Deposita, 2022, 57(8): 1469-1497.
[79]	MIRANDA A C R, BEAUDOIN G, ROTTIER B, et al. Trace element signatures in scheelite associated with various deposit types: a tool for mineral targeting[J]. Journal of Geochemical Exploration, 2024, 266: 107555.
[80]	SCIUBA M, BEAUDOIN G, GRZELA D, et al. Trace element composition of scheelite in orogenic gold deposits[J]. Mineralium Deposita, 2020, 55(6): 1149-1172.
[81]	SCIUBA M, BEAUDOIN G. Texture and trace element composition of rutile in orogenic gold deposits[J]. Economic Geology, 2021, 116(8): 1865-1892.
[82]	YANG C, BEAUDOIN G, SONG Y, et al. Geochemistry of hydrothermal and stream sedimentary rutile in the Tiegelongnan porphyry-epithermal Cu (Au) deposit, Tibet: a tool for exploration[J]. Ore Geology Reviews, 2024, 167: 105970.
[83]	BÉDARD É, DE BRONAC DE VAZELHES V, BEAUDOIN G. Performance of predictive supervised classification models of trace elements in magnetite for mineral exploration[J]. Journal of Geochemical Exploration, 2022, 236: 106959.
[84]	HUANG X W, BEAUDOIN G. Textures and chemical compositions of magnetite from iron oxide copper-gold (IOCG) and Kiruna-type iron oxide-apatite (IOA) deposits and their implications for ore genesis and magnetite classification schemes[J]. Economic Geology, 2019, 114(5): 953-979.
[85]	HUANG X W, BOUTROY É, MAKVANDI S, et al. Trace element composition of iron oxides from IOCG and IOA deposits: relationship to hydrothermal alteration and deposit subtypes[J]. Mineralium Deposita, 2019, 54(4): 525-552.
[86]	HUANG X W, SAPPIN A A, BOUTROY É, et al. Trace element composition of igneous and hydrothermal magnetite from porphyry deposits: relationship to deposit subtypes and magmatic affinity[J]. Economic Geology, 2019, 114(5): 917-952.
[87]	MAKVANDI S, BEAUDOIN G, BETH MCCLENAGHAN M, et al. Geochemistry of magnetite and hematite from unmineralized bedrock and local till at the Kiggavik uranium deposit: implications for sediment provenance[J]. Journal of Geochemical Exploration, 2017, 183: 1-21.
[88]	MAKVANDI S, BEAUDOIN G, BETH MCCLENAGHAN M, et al. PCA of Fe-oxides MLA data as an advanced tool in provenance discrimination and indicator mineral exploration: case study from bedrock and till from the Kiggavik U deposits area (Nunavut, Canada)[J]. Journal of Geochemical Exploration, 2019, 197: 199-211.
[89]	MAKVANDI S, GHASEMZADEH-BARVARZ M, BEAUDOIN G, et al. Principal component analysis of magnetite composition from volcanogenic massive sulfide deposits: case studies from the Izok Lake (Nunavut, Canada) and Halfmile Lake (New Brunswick, Canada) deposits[J]. Ore Geology Reviews, 2016, 72: 60-85.
[90]	MAKVANDI S, GHASEMZADEH-BARVARZ M, BEAUDOIN G, et al. Partial least squares-discriminant analysis of trace element compositions of magnetite from various VMS deposit subtypes: application to mineral exploration[J]. Ore Geology Reviews, 2016, 78: 388-408.
[91]	ZHANG P, ZHANG Z J, YANG J, et al. Machine learning prediction of ore deposit genetic type using magnetite geochemistry[J]. Natural Resources Research, 2023, 32(1): 99-116.
[92]	FRENZEL M. Making sense of mineral trace-element data: how to avoid common pitfalls in statistical analysis and interpretation[J]. Ore Geology Reviews, 2023, 159: 105566.
[93]	ZHONG R C, DENG Y, YU C. Multi-layer perceptron-based tectonic discrimination of basaltic rocks and an application on the Paleoproterozoic Xiong’er volcanic province in the North China Craton[J]. Computers & Geosciences, 2021, 149: 104717.
[94]	LOUPPE G. Understanding random forests: from theory to practice[EB/OL]. (2014-07-28)[2025-02-25]. https://arxiv.org/abs/1407.7502v3.
[95]	郝慧珍, 顾庆, 胡修棉. 基于机器学习的矿物智能识别方法研究进展与展望[J]. 地球科学, 2021, 46(9): 3091-3106.
[96]	LIU Y, SUN T, WU K X, et al. Fractal-based pattern quantification of mineral grains: a case study of Yichun rare-metal granite[J]. Fractal and Fractional, 2024, 8(1): 49.
[97]	王琳, 季晓慧, 杨眉, 等. 基于数据增强和集成学习的矿物图像识别[J]. 地学前缘, 2024, 31(4): 87-94. DOI
[98]	CATÉ A, PEROZZI L, GLOAGUEN E, et al. Machine learning as a tool for geologists[J]. The Leading Edge, 2017, 36(3): 215-219.
[99]	CATÉ A, SCHETSELAAR E, MERCIER-LANGEVIN P, et al. Classification of lithostratigraphic and alteration units from drillhole lithogeochemical data using machine learning: a case study from the Lalor volcanogenic massive sulphide deposit, Snow Lake, Manitoba, Canada[J]. Journal of Geochemical Exploration, 2018, 188: 216-228.
[100]	HILL E J, PEARCE M A, STROMBERG J M. Improving automated geological logging of drill holes by incorporating multiscale spatial methods[J]. Mathematical Geosciences, 2021, 53(1): 21-53.
[101]	陈彦亭, 陈语豪, 王功文, 等. 智能矿山大数据挖掘与知识发现及四维智慧管控: 河北研山露天铁矿为例[J/OL]. 地学前缘, 1-24[2025-04-22]. https://doi.org/10.13745/j.esf.sf.2024.11.63.
[102]	肖凡, 杨华清, 唐奥, 等. 基于机器学习与多源数据融合的东天山戈壁沙漠覆盖区中-酸性侵入岩岩性填图[J]. 地学前缘, 2025, 32(4): 199-212.
[103]	陈国雄, 张越鹏, 罗磊, 等. 数据驱动斑岩型矿床时空预测模型[J]. 地学前缘, 2025, 32(4): 46-59.
[104]	LI W W, HSU C Y. GeoAI for large-scale image analysis and machine vision: recent progress of artificial intelligence in geography[J]. ISPRS International Journal of Geo-Information, 2022, 11(7): 385.
[105]	ZHANG Y F, WEI C, WU S Y, et al. GeoGPT: understanding and processing geospatial tasks through an autonomous GPT[EB/OL]. (2023-07-16)[2025-02-25]. https://arxiv.org/abs/2307.07930v1.
[106]	李三忠, 索艳慧, 戴黎明, 等. 元地球与数字孪生: 思想突破、技术变革与范式转换[J]. 地学前缘, 2024, 31(1): 46-63. DOI
[107]	WANG S Q, HU T, XIAO H, et al. GPT, large language models (LLMs) and generative artificial intelligence (GAI) models in geospatial science: a systematic review[J]. International Journal of Digital Earth, 2024, 17(1): 2353122.

模型名	时间	参数量	研发公司
GPT-3	2020年5月	1 750亿	OpenAI
FLAN	2021年9月	1 370亿	Google
BERT	2019年10月	4 810亿	Google
PaLM	2022年4月	5 400亿	Google
LLaMA	2023年2月	650亿	Facebook
GPT-4	2023年3月	未知	OpenAI
文心一言	2023年4月	未知	百度
MinMax	2023年5月	未知	商汤科技
Gemma7B	2024年2月	93亿	Google
Kimi	2024年10月	2 000亿	Moonshot AI
DeepSeek-R1	2025年1月	6 710亿	深度求索

模型名	时间	参数量	研发公司
GPT-3	2020年5月	1 750亿	OpenAI
FLAN	2021年9月	1 370亿	Google
BERT	2019年10月	4 810亿	Google
PaLM	2022年4月	5 400亿	Google
LLaMA	2023年2月	650亿	Facebook
GPT-4	2023年3月	未知	OpenAI
文心一言	2023年4月	未知	百度
MinMax	2023年5月	未知	商汤科技
Gemma7B	2024年2月	93亿	Google
Kimi	2024年10月	2 000亿	Moonshot AI
DeepSeek-R1	2025年1月	6 710亿	深度求索

对比角度	ChatGPT (GPT-4架构)	DeepSeek (V1架构)
技术架构	密集参数模型	混合专家模型(MoE)
参数量	1.8万亿(估算)	6 850亿
训练成本	>1亿美元	约560万美元
训练方法	三阶段训练(预训练+SFT+RLHF)	五阶段训练(冷启动+推理RL+MoE+蒸馏+场景RL)
推理速度	80 tokens/s	200 tokens/s
样本截止时间	2023年10月	实时更新(周粒度)
核心优势	跨领域泛化能力多模态支持成熟应用程序接口生态	支持本体库嵌入动态计算能耗降多领域数据生成
主要局限	硬件依赖性强知识更新延迟	多模态支持有限领域迁移成本高
典型应用场景	通用对话/跨领域问答/代码生成	专业领域/智能问答/方案优化
部署方案	云端应用程序接口服务	私有化部署/混合云架构

对比角度	ChatGPT (GPT-4架构)	DeepSeek (V1架构)
技术架构	密集参数模型	混合专家模型(MoE)
参数量	1.8万亿(估算)	6 850亿
训练成本	>1亿美元	约560万美元
训练方法	三阶段训练(预训练+SFT+RLHF)	五阶段训练(冷启动+推理RL+MoE+蒸馏+场景RL)
推理速度	80 tokens/s	200 tokens/s
样本截止时间	2023年10月	实时更新(周粒度)
核心优势	跨领域泛化能力多模态支持成熟应用程序接口生态	支持本体库嵌入动态计算能耗降多领域数据生成
主要局限	硬件依赖性强知识更新延迟	多模态支持有限领域迁移成本高
典型应用场景	通用对话/跨领域问答/代码生成	专业领域/智能问答/方案优化
部署方案	云端应用程序接口服务	私有化部署/混合云架构