AI / Deep Learning
AI / Deep Learning: Technology Overview
Artificial intelligence technologies have seen a dramatic increase in public and media attention in recent years. The modern AI boom has been driven mainly by more powerful computers, the increased availability of data for training purposes, and better AI and machine learning algorithms. Today, AI is used in countless applications, including search engines, recommendation systems, targeted advertising, virtual assistants, autonomous vehicles, automatic language translation and facial recognition, to name a few.
Deep learning is a key subfield of machine learning. Deep learning uses complex structures called artificial neural networks that are loosely modelled on the human brain. Information flows through numerous layers of interconnected neurons, where it is processed and evaluated. In essence, AI learns by continuously reassessing its knowledge, forming new connections and prioritising information based on new data encountered. The term ‘deep learning’ refers to the vast number of layers that these networks can utilise. AI powered by deep learning has achieved remarkable advancements, especially in areas such as image and speech recognition.
Generative AI (GenAI) is one of the most powerful forms of AI. Unlike old rule-based AI applications, which could only perform a single task, modern GenAI models are trained on data from many different areas and can handle a wide range of tasks. Due to the vast amount of training data, these models can produce creative outputs. Modern GenAI chatbots, such as ChatGPT and Google Gemini, can generate human-like text and hold conversations on a variety of topics, rather than being confined to a predetermined script. Additionally, these chatbots can produce not only text, but also images, music, and computer code based on their training dataset.
Global trends: sharp rise in AI patents
Global Development of AI patents
AI Patents have risen tenfold over the last decade
The current AI boom is reflected in the significant increase in active AI patents, which increased from around 50,000 to over 1.4 million between 2010 and 2024. Even stronger patent dynamics are evident in the field of deep learning, where the number of active global patents grew from under 13,000 to almost 900,000 during this period. While GenAI currently accounts for only a small proportion of total AI patents (4.2% in 2025), it has experienced extraordinarily high growth in recent years. Active GenAI patents have increased tenfold from around 5,500 in 2020 to almost 56,650 in 2025.
Share of AI of all patent filings over time
Today, every tenth new patent application is related to AI
Even more impressive is the rise of AI when considering only new patent filings. Between 1980 and 2013, the proportion of AI patents within all filings remained consistently below 1%. In 2017, the year of the influential Google paper ‘Attention is All You Need’, which introduced the Transformer architecture that forms the basis of all modern LLMs, the proportion of AI patents increased to nearly 2%. Today, more than one in ten new patent filings are AI-related.
Deep Learning has become the key AI field
Patent development in different AI areas
75% of AI patents are based on Deep learning
Deep learning has become the leading subfield of AI , enabling the development of modern Large Language Models (LLMs) such as OpenAI’s GPT family and Google’s Gemini, which are key examples of Generative AI (GenAI). In 2024, approximately 75% of all active AI patents are in the deep learning field.
Development of deep learning patents in selected applications
Deep learning is transforming various industries
Deep learning is becoming increasingly relevant in the context of mobility, health and industry applications. For example, it is transforming mobility by enabling vehicles to perceive and interpret their surroundings allowing them to detect pedestrians, cyclists and road signs in real time. In healthcare, deep learning is already proving invaluable in diagnostic imaging, where algorithms can detect anomalies in X-rays, MRIs and CT scans with a level of accuracy comparable to or even exceeding that of human experts. In industry, manufacturers use deep learning for quality control, identifying minute defects on assembly lines, and for predictive analytics, optimising energy use and reducing downtime.
Top research countries in deep learning
Active world class patents per inventor country/region
China and the US are far ahead in deep learning
China and the US are engaged in fierce competition in the field of deep learning research. In terms of sheer number of patent publications, China is in a league of its own. However, when it comes to world-class research — the top 10% of global patents with high citations and country coverage — it’s a closely contested race for the top spot.
As the graph on the left shows, the US was clearly ahead for many years, but China has taken the lead since 2021. In 2024, around 43 thousand world-class deep learning patents came from China, compared to around 26 thousand from the US. Every other region lags far behind: Europe has 9,200 world-class patents, Japan has only 2,700, and the rest of the world has 9,200 (mostly South Korea).
Global share of world-class deep learning patents per technology in 2024
EU and Japan only play a substantial role in deep learning in healthcare and smart factories
Examining some of the most significant use cases for deep learning reveals that China and the US are leading the way in world-class research across all major technologies. China is the world leader in patents for deep learning in smart factories, big data, digital agriculture and GenAI. In contrast, the US remains the undisputed leader in applying deep learning to autonomous driving, robotics, the cloud, the metaverse, lidar/radar/sonar, patient data, 3D image modelling, cyber security and biotech, despite China’s higher patent growth rates in almost all areas. The two AI leaders are neck and neck in face recognition, biomedical imaging, speech recognition and AI in fintech.
Notably, the EU accounts for over 10% of global deep learning world-class patents in only three technologies (biotech, patient data and biomedical imaging), while Japan achieves this in just one area (smart factories).
Top research companies in deep learning
Bar Chart Race Deep Learning
Tencent currently holds most active world-class patents in deep learning
Until 2018, US tech companies such as Alphabet, Amazon, Microsoft and Qualcomm dominated deep learning research and held the largest number of world-class patents. Since then, however, certain Chinese companies, first Ping An and later Tencent and Baidu, have climbed to the top positions in terms of active world-class patents.
OpenAI, now synonymous with GenAI following the success of ChatGPT, did not file any patents until early 2023. This reflects its origins as a non-profit organisation dedicated to open research and open-source releases, before shifting to a ‘capped’ for-profit model with Microsoft as a major investor.
