Following the release of OpenAI's ChatGPT, interest in artificial intelligence (AI) has surged. Corporate mentions of AI have increased to more than 715,000 in 2022 from 135,000 in 2020. Our dbDIG proprietary survey revealed that 41% of Americans were familiar with ChatGPT, and nearly 60% reported its adoption in their workplaces. See Figures 1 and 2.

Figure 1: Before today, had you heard of ChatGPT? and Figure 2: My current place of employment...

Source: Deutsche Bank, DBdig

AI is an overnight success that has been many years in the making. To properly anticipate what AI innovation is to come next, we went back to the source, looking at patents and venture capital (VC) deal activity that will translate into the everyday AI applications of the future.

Companies are set to benefit from the explosion of AI innovation research is now being deployed in the commercial world after many years of being contained to academia. Academia is usually the first mover: there is generally a 15-year lag from publication to patent. Published patents – similar to VC money – tend to around a one to three-year lead-time to production.

Using the World Intellectual Property Organisation’s (WIPO) Patentscope tool, we collected more than 175,000 published AI patent entries from 2012 to 2022 across 193 WIPO members, broken down into five broad categories. For VC investment, we used the OECD.AI database covering over 24,000 deals across 92 economies from 2012 to 2022. This article summarises our main findings.

Over the past decade, around 80% of AI-related venture capital (VC) deals focused on sectoral applications. Another 8% were in autonomous machines and vehicles, while 13% were in the semiconductor space. The distribution of AI-related patents is similar, with almost two-thirds in sectoral applications, 19% in horizontal technology solutions such as developer tools and computer vision, 16% in autonomous machines and vehicles, and 4% in semiconductors.

Figure 3: Number of AI-related VC deals 2012-2022 and Figure 4: AI-related patents published 2012-2022

Source: Deutsche Bank, OECD.AI

Patent growth areas

The application of AI across sectors is experiencing remarkable growth. VC deals related to AI have increased tenfold from 2012 to 2022, reaching a total deal value of US$62bn in 2022. Patents published within sectors such as consumer, industrial, IT, transportation, healthcare, and financial services have also seen a six-fold increase since 2012. Notably, established technology giants like IBM, Samsung, Intel, LG Electronics, and Qualcomm hold significant AI-related patents.

VC investment in AI has been concentrated in healthcare, IT, and financial services. In healthcare, AI is revolutionising medicine, clinical decision support, genetic analytics, healthcare administration, and personal health. The European Commission predicts a cumulative impact of €105bn by 2030. In IT, AI optimises traditional IT functions, while in financial services, it enhances advanced analytics, process automation, robo-advisers, and self-learning programs.

Patent activity is prominent in the industrial, consumer, and transportation sectors. Industrial AI automates processes and unlocks industrial data for improved efficiency, this is projected to have a cumulative impact of €200bn n by 2030. Consumer AI affects business models and specific categories such as media, advertising, gaming, education, price optimisation, and e-commerce recommendations. In transportation, AI assists with logistics and fleet management, exemplified by WiseTech Global's automation software for shipping logistics.

Autonomous machines and semiconductors

Autonomous machines, including self-driving cars and intelligent robots, have garnered significant investment and attention. They represented 8% of VC deals and 16% of patents in the past decade. Tesla's success with advanced autonomous vehicle systems, such as autopilot and self-driving features, using advanced cameras and neural networks, played a crucial role. VC deals in autonomous machines increased from 22 to 363 from 2012 to 2022, with a total value of US$11.7bn in 2022. Patents in this category grew seventeen-fold since 2012. Intelligent robotics, operating with minimal human intervention, rely on AI for training and adaptation, with companies like Samsung, LG Electronics, Intel, SZ DJI, and Beijing Didi Infinity Tech holding significant patents. The rise in VC deals and patents indicates broader industry interest and investment in the development of autonomous vehicle technologies beyond Tesla.

Semiconductors play a vital role in the AI landscape, with 13% of AI-related VC deals and 4% of patents. VC deals in semiconductors rose from 40 to 515 from 2012 to 2022, with a total value of US$9.4bn in 2022. Patents in this field increased seven-fold. AI enhances processor design, improving efficiency and speed with innovations like low-precision calculations. The US leads in upstream design, while China excels in assembly and testing. Edge AI focuses on compressing algorithms for semiconductor environments, and intelligent sensors and devices provide real-world measurements. Integrating AI into these subcategories drives technological advancements.

Generative audio, powered by AI like ChatGPT, has made remarkable progress in creating realistic voices and synthetic audio from text. Corporate documents mentioning "generative audio" have increased more than thirteen-fold from Q1 2020 to Q4 2022.

This technology democratises music and sound development, removing the need for specialised professionals. It breaks free from the limitations of voice actors and musicians, opening possibilities in gaming, communications, music, news, and healthcare.

Major players like Sony, Amazon, Huawei, Bytedance, Adobe, Apple, and Tencent hold significant patents in this field. The four primary categories include synthetic voice, speech interaction, music generation and audio editing.

In summary, the past decade witnessed a significant surge in AI innovation, with more than tenfold growth in VC deals to 3,884 in 2022, totalling 24,310 deals. The value of these deals rose from US$1.8bn to US$83bn. Published patents increased seven-fold to 32,800. Most AI innovation focused on transportation, industry, and consumer applications. Generative audio innovation is expected to transform gaming and film production.

Challenges include market concentration, data accuracy, transparency, and varying regulatory frameworks. Cybersecurity is a growing concern with AI attacks. Robust regulations and cybersecurity measures are crucial for ethical and secure AI innovation.

Marion Laboure is a Senior Strategist at Deutsche Bank Research and a Lecturer at Harvard University. This article is based on the Deutsche Bank Research study, AI in Action: Where is the smart money going? (May 2023)