本研究对专利活动数据进行统计分析。引言提供了有关专利研究的具体情况、研究领域和方法的简短概念。主要部分概述了主要地质调查类别中的一般统计数据、分布、动态图、人工智能发明的份额，对增长、增长率的分析，以及为用户和技术开发人员提供的一些可能的结论。

专利研究的具体情况
专利活动的研究依赖于专利局的官方出版物，例如授权专利和专利申请，它们提供了非常有价值的信息。专利文献有以下主要优点：

• 实际相关性——与理论或假设的解决方案不同，专利文献侧重于对实际利用有价值的发明。这种对实用性的重视确保了所获得的信息可以直接应用于现实世界的场景。
• 及时性——现有申请通常在提交后大约 18 个月内发布，这意味着在产品或技术引入市场之前信息就已可用。专利信息的这种最新性使研究人员能够保持领先地位。
• 全面洞察力的文件不仅提供了有关技术本身的详细信息，还披露了有关谁申请专利以及何时申请的关键数据。这些附加信息增强了对技术前景的理解并有助于进一步分析。

因此，专利信息揭示了对特定领域内正在进行的项目和新兴技术的非常有价值的见解。
虽然一些商业服务提供出于研究目的访问专利公布数据库的权限，但也有强大的免费服务，包括世界知识产权组织的 Patentscope、欧洲专利局的 Espacenet 和 Google Patents。

本研究使用谷歌专利数据库。Google Patents 数据库中可访问的专利文献总数超过 105 个国家/地区的 1900 万份官方出版物。

研究领域和方法
本研究重点检索和分析地质调查六大类中人工智能技术的专利活动，涵盖1980年1月1日至2023年5月1日。研究起点与核心技术的出现时间一致。讨论人工智能工具在实际计算中的潜在用途的科学出版物。

该研究的地质调查类别如下：

• 地震和声学
• 磁、电、电磁和核磁共振波谱 (NMR)
• 核辐射，包括伽马射线
• 光学的
• 引力
• 方法和地理建模的结合，与特定测量无关

搜索查询具有以下一般结构，其中第一部分是恒定的，旨在获得在地质测量中使用人工智能的主要领域的结果，第二部分旨在过滤特定地质测量类别的结果。
由于与人工智能相关的术语尚无统一的术语，查询中包括“人工智能”、“神经网络”、“机器学习方法”等常见关键词以及近年来为人工智能建立的具体合作专利分类类别。人工智能。每个类别都需要详细的查询阐述，平均有 1,900 个符号、160 个逻辑运算符和 120 个括号。

专利文件的公布有延迟。根据去年的统计，只有一半的发明是在申请后的17至29个月内发表的。考虑到这一点，2021 年的数据进行了乘以 2 的调整。

搜索结果处理产生了专利活动的概览，说明了人工智能在地质调查中的应用不断发展。

人工智能地质调查发明综合统计

所有六大类的发明总数为 16,187 项，其中 730 项涉及人工智能。

最早实施人工智能的发明属于第 6 类，“方法和地理建模的结合”。“用于产生特别是钻孔穿过的地质构造相的记录特征的方法”（法国专利 2,520,882）的发明者是塞拉·奥博托、阿伯特·海登、科巴特·伊夫和文森特·菲利普。Societate de Prospection 脡lectrique（现 SLB）于 1982 年 2 月 2 日提交了 40 多年前的专利申请。这是 1982 年唯一将人工智能融入地质测量的发明。该发明使用带有人工智能元素的软件来提供关于分析区间内地层性质的结论。

如图1所示，未来 33 年仅提交了 85 件涉及人工智能的专利申请。然而，自2016年以来，发明数量开始急剧增加。2016年，发明创造38项，增长率高达533.3%。2017年，发明数量稳定但随后继续增加。值得注意的是，仅 2019 年就提交了 94 项包含人工智能的专利申请，超过了最初 33 年期间提交的总数。2018年至2020年间的增长率为每年66.7%至93.6%。2021年增速放缓至28.6%。

从 1980 年到 2021 年，地质调查领域的发明总数（无论是否包含人工智能）均有所增长（见图 1 中的灰线），这表明地质调查发明总投资的增长。

人工智能发明的动态增长表明，在 2010 年代后半叶，开发人员越来越了解人工智能作为地质调查工具的价值，并开始系统地提取这一价值。未来几年，将人工智能融入地质测量的发明的增长率可能会稳定下来。此外，利用人工智能开发地质测量技术的活动水平在不同类别之间存在显着差异。

人工智能发明按类别分布

图 2是一个饼图，显示了不同类别中包含人工智能的发明的分布情况。第 1 类“地震和声学”占发明的大部分，占发明总数的 64%。这种浓度可能归因于以下因素：

• 地震方法的广泛使用
• 非常完善的数据收集技术
• 数据流动频率低，数据处理方便

第 6 类“组合和地理模型”占发明的第二大份额，占所有发明的四分之一。此类发明的大量发明可归因于这样一个事实：传统上，经验丰富的地质学家是唯一能够结合不同数据并创建地质模型的人。但现代人工智能工具正在加速、简化并提高此类流程的准确性。
其余类别总共仅占发明的 11%，其中第 2 类“磁、电、电磁、核磁共振”占总数的 6%。尽管与地震方法相比，这些方法收集的数据较少，但跨类别的分布将来可能会发生变化。检查每个类别的动态图可以提供进一步的见解。

按年份和类别划分的动态

图 3显示了 2020 年至 2021 年所有涵盖类别的专利活动增长情况。值得注意的是，类别 1“地震和声学”经历了最高的激增，2020 年增加了 57 项发明，2021 年增加了 26 项发明。如上图所示，与其他类别相比，这种大幅增长促成了目前第一类发明的主导地位。

然而，为了对未来的发明活动做出假设，不仅要考虑绝对增长率，还要考虑近年来的增长率，这一点至关重要，如图4 所示。

图4显示，2020年，第一类尽管拥有大部分累计发明创造，但增长率排名第四。到2021年，跌至最后一位，增长率仅为21%。同样，累计发明排名第二的第6类“组合和地理模型”的增长率也出现下滑，2021年排名倒数第二。

相比之下，累计发明数量最少的第5类“重力”，在2021年却呈现出惊人的300%的增长率。此外，第2类“磁、电、电磁、核磁共振”；第 3 类，“辐射”；第四类“光学”在2021年的增长率均为100%。

2-5类中观察到的趋势可能归因于低基数效应的影响。然而，在未来几年中，这些类别的显着增长率很可能会导致人工智能发明在不同类别中的分布更加均匀。

影响每个类别中人工智能发明的增长率和数量的因素之一是分配给人工智能开发技术的预算在研发总预算中所占的份额。可以通过计算特定类别的人工智能发明占总发明的比例来评估份额。

各类别人工智能发明占比
综合统计部分的数据显示，目前人工智能累计发明占比仅为4.51%。然而，这个数字是扭曲的，因为历史数据反映了早期人工智能的使用很少。

图 5显示了所有类别中纳入人工智能的发明所占份额的增长趋势。2016年，这一比例为0%至6.9%。到 2021 年，这一比例将增至 6.3% 至 22.2%。值得注意的是，在专利活动最高的类别中，即类别 1“地震和声学”和类别 6“组合和地理模型”，包含人工智能的发明比例达到约 14%。这表明技术开发人员正在将越来越多的开发组合分配给人工智能驱动的工具。

这些趋势表明，未来几年，地质测量领域纳入人工智能的发明数量可能会增加。这种增长可归因于两个因素：

• 如一般统计部分所示，地质测量发明总投资增加
• 分配给人工智能技术开发的预算份额越来越大

对人工智能发明的分布、动态和份额的分析得出了一些总体结论。
针对用户和技术开发人员的一般结论 与
用户和开发人员相关的结论将根据他们的具体需求、市场地位和资源而有所不同。然而，可以得出一些一般性结论。

由于第 1 类“地震和声学”和第 6 类“组合和地质模型”积累了最多的发明，因此用户可以预期这些类别中的新产品将融入人工智能工具。这些类别中不使用人工智能的产品可能看起来已经过时。虽然其他类别中不使用人工智能的产品可能会在一段时间内继续保持其市场地位，但预计大多数地质测量产品最终将采用人工智能工具进行数据分析。

专注于地震和声学测量或地理建模产品的开发人员应该使用人工智能工具来保持竞争力。然而，由于激烈的竞争和专利侵权的风险，鉴于该类别有数百项有效专利，他们也可能面临挑战。除了在技术挑战上花费资源外，他们还可能被迫花费额外的资源来克服法律问题。相比之下，其他四个类别的产品开发商到目前为止可能还没有遇到这些困难。

地质调查技术中人工智能的应用正在不断发展。使用各种研究方法（包括专利研究）监测该领域的趋势可以为产品开发和使用的决策提供有价值的见解。

未来石油和天然气勘探的专利研究可以集中于检索人工智能在地质调查方法特定领域的实施信息。或者，它可以探索其他现代工具的使用，例如量子计算，这些工具已经在该领域获得了专利。

This research conducts a statistical analysis of patent activity data. The introduction provides short notions on specifics of patent research, fields of research, and methodology. The main section provides an overview of the general statistics, distribution, dynamics diagrams, the share of inventions with AI in the main geological survey categories, an analysis of growth, growth rate, and some possible conclusions for users and technology developers.

Specifics of Patent Research
Research on patent activity relies on official publications from patent offices, such as granted patents and patent applications, which provide highly valuable information. Patent documentation has the following main advantages:

• Practical relevance—Unlike theoretical or hypothetical solutions, patent documentation focuses on inventions that are valuable for practical exploitation. This emphasis on practicality ensures that the obtained information is directly applicable to real-world scenarios.
• Timeliness—Patent applications typically are published approximately 18 months after filing, which means the information becomes available before the products or technologies are introduced in the market. This up-to-date nature of patent information allows researchers to stay ahead of developments.
• Comprehensive insight—Patent documents not only provide detailed information about the technology itself but also disclose crucial data on who filed the patent and when. This additional information enhances the understanding of the technology landscape and facilitates further analysis.

Thus, patent information unveils very valuable insights into ongoing projects and emerging technologies within a specific field.
While several commercial services offer access to patent publication databases for research purposes, there are also powerful free services, including Patentscope from the World Intellectual Property Organization, Espacenet from the European Patent Office, and Google Patents.

The presented research uses the Google Patents database. The total accessible number of patent documents in the Google Patents database exceeds 19 million official publications from 105 countries.

Field of Research and Methodology
This study focuses on retrieving and analyzing patent activity for technologies utilizing AI within six main categories of geological surveying and covers the period from 1 January 1980 to 1 May 2023. The research starting point coincides with the time when there were core scientific publications discussing the potential use of AI tools for practical computations.

The study’s geological surveying categories are the following:

• Seismic and acoustic
• Magnetic, electrical, electromagnetic, and nuclear magnetic resonance spectroscopy (NMR)
• Nuclear radiation, including gamma ray
• Optical
• Gravitational
• Combination of methods and geomodelling, not related to particular measurements

The search queries have the following general structure where the first part is constant and aims to get results in the main field of using AI in geological surveying and the second part aims to filter results for specific geological surveying categories.
Because there is no consistent terminology related to AI, the query included common keywords such as “artificial intelligence,” “neural network,” “machine learning methods,” and specific cooperative patent classification classes established in recent years for AI. Each category required a detailed query elaboration, with an average of 1,900 symbols, 160 logic operators, and 120 brackets.

Patent documents are published with a delay. Based on previous year’s statistics, only half of the inventions were published within a period of 17–29 months from filing. To account for this, the data for 2021 was adjusted by multiplying it by 2.

The search result processing yielded an overview of patent activity, illustrating the continuous development of AI applications in geological surveys.

General Statistics for Inventions Incorporating AI in Geological Surveying

The total number of inventions in all six categories is 16,187, of which 730 incorporate AI.

The earliest invention implementing AI is in Category 6, “Combination of Methods and Geomodelling.” The “method for producing a recording characteristic notably of the facies of geological formations traversed by a borehole” (French patent 2,520,882) was invented by Serra Oberto, Abbott Hayden, Kerbart Yves, and Vincent Philippe. The Société de Prospection Électrique (now SLB) filed the patent application more than 40 years ago on 2 February 1982. This was the only invention that incorporated AI in geological surveying in 1982. The invention uses software with AI elements to provide conclusions about the nature of the formations in the analyzed interval.

As Fig. 1 shows, only 85 patent applications involving AI were filed for the next 33 years. Since 2016, however, the number of inventions has started to increase sharply. In 2016, there were 38 inventions, marking a staggering growth rate of 533.3%. In 2017, the number of inventions stabilized but then continued to increase. Notably, in 2019 alone, 94 patent applications incorporating AI were filed, surpassing the total number filed during the initial 33-year period. The growth rate between 2018 and 2020 ranged from 66.7% to 93.6% per year. In 2021, the growth rate slowed down to 28.6%.

The overall number of inventions in the field of geological surveys, both with and without AI, grew from 1980 to 2021 (see the gray line in Fig. 1), which indicates growth in the total investment in inventions for geological surveys.

The dynamic growth of inventions incorporating AI suggests that, in the second half of the 2010s, developers increasingly understood the value of AI as a tool in geological surveys and began systematically extracting this value. The growth rate of inventions incorporating AI in geological surveying may stabilize in the coming years. Furthermore, the activity levels in developing geological surveying technologies with AI vary significantly across different categories.

Distribution of Inventions Incorporating AI by Categories

Fig. 2 is a pie chart illustrating the distribution of inventions incorporating AI across different categories. Category 1, “Seismic and Acoustic,” constitutes the majority of inventions, representing 64% of the total inventions. This concentration might be attributed to the following factors:

• Widespread use of seismic methods
• Very well established data collection technologies
• Convenience in data processing because of the low frequency of data flow

Category 6, “Combination and Geomodels,” accounts for the second-largest share of inventions, at a quarter of all inventions. The significant number of inventions in this category can be attributed to the fact that, traditionally, highly experienced geologists were the only ones able to combine different data and create geomodels. But modern AI tools are speeding up, simplifying, and improving the accuracy of the processes in this category.
The remaining categories collectively account for only 11% of inventions, with Category 2, “Magnetic, Electrical, Electromagnetic, NMR,” representing 6% of the total. Although these methods collect less data compared to seismic methods, the distribution across categories may evolve in the future. Examining the dynamic diagrams for each category provides further insight.

Dynamics by Years and Categories

Fig. 3 shows the growth in patent activity in all covered categories from 2020 to 2021. Notably, Category 1, “Seismic and Acoustic,” experienced the highest surge, with an increase of 57 inventions in 2020 and 26 inventions in 2021. This substantial growth has contributed to the current dominance of inventions in Category 1 compared with other categories, as shown in the previous diagrams.

To make assumptions about future invention activity, however, it is crucial to consider not only the absolute growth but also the growth rate in recent years, as depicted in Fig. 4.

Fig. 4 shows that, in 2020, Category 1, despite having the majority of accumulated inventions, ranked fourth in terms of growth rate. By 2021, it dropped to the last position, with a growth rate of only 21%. Similarly, Category 6, “Combination and Geomodels,” which ranks second in accumulated inventions, experienced a decline in growth rate and ranked second-to-last in 2021.

In contrast, Category 5, “Gravity,” with the fewest accumulated inventions, exhibited a remarkable 300% growth rate in 2021. Furthermore, Category 2, “Magnetic, Electrical, Electromagnetic, NMR”; Category 3, “Radiation”; and Category 4, “Optical,” all had a growth rate of 100% in 2021.

The observed trend in Categories 2–5 could be attributed to the influence of the low base effect. It is highly likely, however, that, in the upcoming years, the significant growth rate in these categories might lead to a more even distribution of inventions incorporating AI across the different categories.

One of the factors contributing to the growth rate and the number of inventions incorporating AI in each category is the share of the budget allocated to developing technologies with AI within the total research and development budget. The share can be evaluated by calculating the proportion of inventions with AI among the total inventions in a specific category.

Share of Inventions Incorporating AI in Each Category
The data presented in the General Statistics section reveals that the share of accumulated inventions incorporating AI is currently only 4.51%. This figure is distorted, however, because of historical data reflecting a minimal use of AI in the early years.

Fig. 5 illustrates the growing trend in the share of inventions incorporating AI across all categories. In 2016, this share ranged from 0% to 6.9%. By 2021, it increased to between 6.3% and 22.2%. Notably, in the categories with the highest patent activity, namely Category 1, “Seismic and Acoustic,” and Category 6, “Combination and Geomodels,” the share of inventions incorporating AI reached approximately 14%. This indicates that technology developers are allocating an increasing portion of their development portfolio to AI-powered tools.

The trends suggest a potential increase in the number of inventions incorporating AI in the field of geological surveying in the coming years. This growth could be attributed to two factors:

• An increase in total investment in inventions for geological surveying, as indicated in the General Statistics section
• The growing share of the budget allocated to developing technologies with AI

The analysis of distribution, dynamics, and share of inventions incorporating AI suggests some overarching conclusions.
General Conclusions for Users and Technology Developers
The conclusions relevant to users and developers will vary depending on their specific needs, market position, and resources. Some general conclusions can be made, however.

Because Category 1, “Seismic and Acoustic,” and Category 6, “Combination and Geomodels,” have accumulated the greatest number of inventions, users can expect that new products in these categories will incorporate artificial intelligence tools. Products within these categories that do not use AI might seem obsolete. While products that do not use AI in other categories may continue to maintain their market position for a certain period, it is anticipated that the majority of products for geological surveying eventually will incorporate AI tools for data analysis.

Developers focusing on products for seismic and acoustic surveying or geomodelling should use AI tools to be competitive. They may also face challenges, however, because of intense competition and the risk of patent infringement, given the hundreds of valid patents in this category. Besides spending resources on technological challenges, they could be forced to spend additional resources to overcome legal issues. In contrast, it is likely that developers of products in the other four categories have not encountered these difficulties thus far.

The landscape of AI adoption in geological survey technologies is continuously evolving. Monitoring trends in this area using various research methods, including patent research, can provide valuable insights for decision-making in product development and use.

Future patent research in oil and gas exploration can be focused on retrieving information about the implementation of AI in specific areas of geological survey methods. Alternatively, it can explore the use of other modern tools, such as quantum computing, which have already been patented in this field.