Global trends
GenAI is expected to have a huge impact across many industries as it finds its way into products, services and processes, becoming a technological enabler for content creation and productivity improvement. For example, a recent study by McKinsey estimates that GenAI could add between US$2.6 trillion and US$4.4 trillion annually across a wide range of industry use cases (McKinsey 2023). The firm believes that banking, high tech and life sciences are among the industries that could see the greatest impact from GenAI.
Based on our analysis of GenAI patents, we have identified the applications where research activities are focused on. The following list shows the 21 application areas identified, ranked according to the number of published patent families within the last decade. A short description of current GenAI trends within these applications including patent examples is included in the Appendices.
Software and other applications
Life sciences
Document management and publishing
Business solutions
Industry and manufacturing
Transportation
Security
Telecommunications
Personal devices
Banking and finance
Physical sciences and engineering
Education
Entertainment
Arts and humanities
Computing in Government
Networks and smart cities
Industrial property, law, social and behavioral sciences
Cartography
Military
Energy management
Agriculture
The largest application domain is the software category. However, to note, a large number of patent families cannot be assigned to a specific application and are instead included in the category software/other applications.
Patent families in the other categories are smaller in number, with life sciences in second place (5,346 patent families between 2014 and 2023) and document management and publishing (4,976) in third place (Figure 26). Other notable applications with GenAI patent families ranging from around 2,000 to around 5,000 over the same period are business solutions, industry and manufacturing, transportation, security and telecommunications.
In general, patent growth has been high in all applications since 2014. However, over the last three years there have been diverging growth trends with very high growth rates in both smaller application areas such as agriculture and energy management and large fields such as life sciences, security and physical sciences/engineering. In contrast, the number of patent families has stagnated or even declined in certain application areas such as telecommunications, military, arts and humanities or industrial property/law/social and behavioral sciences (Figure 27).
Top patent owners
Tencent is the top company in several GenAI applications. The social media/gaming company leads in software/other applications, document management and publishing, personal devices, entertainment, security and arts and humanities (Table 10).
Ping An Insurance Group ranks second overall. It has transformed itself from a traditional financial services company into a technology ecosystem with several technology subsidiaries in various industries. The company holds the top spot in the global ranking in business solutions, life and medical sciences, banking and finance, computing in government, industrial property/law/social and behavioral sciences, education and networks and smart cities.
Baidu is the leader in physical sciences and engineering and arts and humanities and also a key player in software and other applications, document management and publishing, and transportation.
IBM has a strong research position in various fields such as software/other applications, document management and publishing, business solutions as well as life and medical sciences.
Alibaba ranks fifth overall, with particularly strong research priorities in software/other applications, document management and publishing, business solutions and arts/humanities.
Samsung is ranked sixth overall, with a research focus on areas such as telecommunications (world leader in patents) and personal devices (third place)
Alphabet/Google’s strengths in GenAI are in software/other applications, life and medical sciences, transportation and telecommunications. The company is also the world leader in GenAI in agriculture.
The majority of Bytedance’s GenAI patents are in the fields of software/other applications and document management and publishing.
Microsoft has developed many GenAI patent families over the last decade in software/other applications, document management and publishing, business solutions and personal devices.
BBK Electronics has a clear research focus on software/other applications.
Other notable GenAI research companies outside the top 10 include:
Huawei (China): the Chinese ICT company has a research focus on software/other applications, document management and publishing, business solutions, telecommunications and transportation.
Adobe (US): the software company has many GenAI patent families in the areas of software/other applications and document management and publishing.
Bosch (Germany): the world’s biggest car supplier is a top GenAI research company in transportation.
Netease (China): the Chinese gaming company is the world leader in GenAI patents in entertainment.
Siemens (Germany): the German industrial conglomerate has developed the second most GenAI patents in life and medical sciences.
Nvidia (US): the leading chipmaker for high-tech AI chips has many GenAI patents in the fields of software/other applications, entertainment and transportation.
Sony Group (Japan): the Japanese conglomerate has the third most GenAI patents in entertainment.
LG Electronics (Republic of Korea): the electronics conglomerate from the Republic of Korea is in second place in GenAI in both transportation and networks/smart cities.
Bank of China (China): the large Chinese bank has many GenAI patents in banking/finance (second place).
UiPath (US): the robotic process automation company has developed most GenAI patent families in industry and manufacturing and is also a research leader in business solutions.
Hitachi (Japan): the Japanese conglomerate has the second most GenAI patent families in energy management.
Autodesk (US): the software company is one of the top research players in GenAI in physical sciences and engineering as well as industry/manufacturing.
State Grid (China): the world’s largest utility is the world leader in GenAI patents in energy management and has also developed many GenAI patents in the areas of industry and manufacturing and security.
Among the top research organizations, the Chinese Academy of Sciences (CAS) stands out. It holds the largest number of GenAI patent families in many applications such as software/other applications, life/medical sciences and document management and publishing (Table 11).
The Chinese Tsinghua University has a strong position in software/other applications, life/medical sciences, document management and publishing, and transportation. It also ranks first among research organizations in industrial property/law/social and behavioral sciences.
The Chinese Zhejiang University has developed many GenAI patent families in software/other applications, document management and publishing, transportation and security.
The Zhejiang University of Technology from China is ranked fourth overall, excelling in fields as diverse as life and medical sciences, security, transportation and telecommunications.
The National Research Council of Science and Technology from the Republic of Korea has a research focus on GenAI in software/other applications.
Key locations of inventors
The category software/other applications is the dominant GenAI research field for all top inventor locations in terms of patent families between 2014 and 2023 (Table 12).
China is the leading inventor location for all GenAI applications. China’s relative lead is particularly pronounced in fields such as software/other applications, document management and publishing, banking and finance, energy management, cartography and industrial property, legal, social and behavioral sciences.
The US is in second place and has a very high share of all GenAI patent families in physical sciences and engineering, life and medical sciences, military, agriculture, entertainment and education.
The Republic of Korea shows a relatively high number of GenAI patent families in business solutions, education and agriculture. In relative terms, Japan has a strong research position in entertainment, and arts and humanities. India has an above-average share of all GenAI patent families in networks and smart cities. The UK stands out in physical sciences and engineering. Germany is in a good research position in physical sciences and engineering and industry and manufacturing.
Connection between core models and applications
There is an interdependence between GenAI models and applications (Figure 28):
GAN models play a dominant role in almost all GenAI applications, with the most pronounced relevance in transport. GAN models are particularly useful for generating images and videos. Therefore, these models play an important role in areas such as autonomous driving. For example, GAN models can be used to generate images or videos to train autonomous vehicles.
VAEs are important for application areas such as physical sciences and engineering, networks/smart cities or personal devices. VAEs excel at capturing the underlying structure (latent space) of the data they are trained on. They can be used to generate new data such as images or to detect anomalies in data points. VAEs are a promising tool for various tasks in physical sciences and engineering, particularly in areas of data exploration or the discovery of novel materials, molecules and designs. They can also contribute to the generation of digital twins.(1)A digital twin is a virtual replica of a physical object or system. It is a computer-generated counterpart that mirrors the real world in real time, allowing you to monitor its performance, predict issues and optimize its operation.
LLM models have a particularly high share in applications such as document management and publishing, business solutions, personal devices, education, entertainment or industrial property, law, social and behavioral sciences. LLMs excel at text-based tasks such as generating creative text formats which can be used for content generation, machine translation or chatbots. However, new multimodal LLMs are also able to process other data such as images, videos or code.
Diffusion models excel at tasks requiring high-fidelity and controllable output generation. The patent data analysis shows a high share of diffusion models in computing in government, networks and smart cities, life and medical sciences and business solutions. For example, diffusion models can be used in the area of life and medical sciences to manipulate medical images while preserving key information or to generate realistic protein sequences and structures.
Autoregressive models are applicable to various tasks in GenAI, particularly those that involve sequential data generation. They are particularly important in the field of banking and finance due to their strengths in sequential data analysis. This is useful for forecasting financial data such as stock prices, exchange rates, interest rates, etc., as well as for tasks such as fraud detection or credit risk assessment.
Connection between modes and applications
Regarding the use of GenAI modes in GenAI applications, we can see the following results (Figure 29):
Image, video is the dominant data type for GenAI in software and other applications. It is also a very important GenAI data source for transportation, life and medical sciences, arts and humanities, agriculture and cartography.
Text data plays a major role for GenAI patents in software and other applications, document management and publishing, arts/humanities, industrial property law/social and behavioral sciences.
Speech/voice/music is the most important data type in personal devices, telecommunications, personal devices and education and also relevant for software and other applications.
Molecules/genes/proteins data is mainly used for GenAI in life and medical sciences as well as physical sciences and engineering.
Software and code plays an important role for GenAI patent families in security and banking/finance.
Other modes are an important source of data for GenAI in software/other applications, banking and finance, business solutions and industry/manufacturing.