Chinese universities aid huge surge in AI inventions

A new study by the World Intellectual Property Organization (WIPO) in Geneva, Switzerland, has documented “a massive recent surge” in artificial intelligence-based inventions, with China-based universities making a strong showing as top higher education and scientific research institutions in this field, which is generally dominated by United States and Japanese companies.

Companies represent 26 out of the top 30 artificial intelligence (AI) patent applicants, while only four are universities or public research organisations.

The international agency’s first report specifically on AI, WIPO Technology Trends 2019: Artificial intelligence, includes a ranking of global institutions that filed the most patent applications for AI inventions, which often comprise several patents and are called a patent family.

Some 167 universities and public research organisations are ranked among the top 500 patent applicants – 110 Chinese, 20 from the US, 19 from South Korea, four from Japan and four European public research organisations.

The WIPO report released on 31 January indicated that while AI patent activity is dominated by US and Japanese companies – huge conglomerates that file patents in multiple sectors – followed by South Korea’s Samsung, universities contribute significantly to research in specific AI fields such as distributed AI, some machine-learning techniques and neuroscience and neurorobotics.

And Chinese universities such as Xidian University in Xi’an, with large numbers of patent families in telecommunications, and Zhejiang University, with its AI patents for industry and manufacturing, are among the top 30 AI patent filers, both companies as well as university and research organisations. Beijing’s Tsinghua University is also a key player.

This compares to top US universities and public research organisations such as the University of California, Massachusetts Institute of Technology and Columbia University, with numerous AI patent applications in life and medical sciences as well as physical sciences and engineering.

The top two Japanese research organisations – the National Institute of Advanced Industrial Science and Technology and the National Institute of Information and Communications Technology – also have strong portfolios in life sciences and telecommunications but also in arts and humanities, “a pattern that is unique among the top universities and public research organisations”, according to the WIPO report.

Life and medical sciences are also prominent in AI patents in Europe, including in the United Kingdom and Germany.

South Korea’s Electronics and Telecommunications Research Institute stands out as second in patent filing among universities and public research organisations, ranking among the top 30 patent applicants overall.

Chinese universities’ strong showing

Chinese organisations make up 17 of the top 20 academic players for AI patents and 10 of the top 20 in AI-related scientific publications.

Chinese organisations are particularly strong in the emerging technique of deep learning, with the Chinese Academy of Sciences (CAS) on top with more than 2,500 patent families and more than 20,000 scientific papers published on AI, the report said. Moreover, CAS has the largest deep learning portfolio, with 235 patent families.

Song Hefa, deputy dean of the Intellectual Property School at CAS, said one of the major reasons for CAS’s success in AI patents has been “vigorously carrying out IP [intellectual property] training and information. Since 2008, 16,000 people have been trained and at the end of 2016, CAS had 1,891 people engaged in IP management, transfer and service.”

Overall, “Chinese organisations are consolidating their lead, with patent filings having grown on average by more than 20% per year from 2013 to 2016, matching or beating the growth rates of organisations from most other countries,” the WIPO report said.

Boi Faltings, professor of AI at École Polytechnique Fédérale de Lausanne (EPFL), Switzerland, said: “Since AI-related innovations are enabled by data, the organisations that generate the most AI-related patents are often the ones that own the most data.” This explained “in particular the surprisingly strong position of China”, he added.

Faltings noted in the report that there are “far fewer obstacles to collecting vast amounts of data in China than in other countries” and “China has the best training data collections for speech recognition, human behaviour modelling and medical data”.

“Research in AI technologies in Chinese universities is now catching up with their top-class peers worldwide. For some AI techniques, the gap between Chinese universities and research institutions and other AI giants is becoming smaller and smaller; and for some functional applications, we have made great progress and may be a few steps ahead of other key players,” said Haifeng Wang, senior vice president at Chinese internet company Baidu.

“Behind these results is a continuous growth of the AI talent pool in China, thanks to the efforts of AI-related enterprises, such as Baidu, working with Chinese universities to accelerate technological innovation and talent cultivation,” Wang was quoted in the report as saying.

AI scientific productivity

Since AI emerged as a field of research in the 1950s, AI patent applicants reached nearly 340,000, with more than 1.6 million scientific publications produced. Notably, AI-related patenting is growing rapidly, with more than half of identified inventions published since 2013.

However, the ratio of scientific papers to inventions has decreased from 8:1 in 2010 to 3:1 in 2016 – indicative of a shift from theoretical research to the use of AI technologies in commercial products and services.

The WIPO report also indicated that some areas of AI are growing more quickly than others. For example, machine learning is the dominant AI technique disclosed in patents and is included in more than one-third of all identified inventions.

“It is clear that we are going through a renaissance of AI and this is also reflected in the increasing numbers of scientific publications and patent applications, with half of the patenting activity in AI having taken place since 2013,” the WIPO report said.

“This boom in AI patenting activity, combined with growth rates across different AI technologies over the period 2013-2016, which are much higher compared with those observed in general in all fields of technology, clearly indicate that this trend can be expected to continue,” it said.

Hamza Alfrmawi, ICT expert at the Islamic Development Bank Alumni and Science Development Network, told University World News that the WIPO report indicates the start of the harvesting process, picking up the fruits of AI research through moving from the research phase in the lab to the production of AI technologies in commercial products, services and processes.

But it also “sends a warning sign to universities in developing countries, especially African and Middle East countries, to quickly start launching AI educational and research programmes to fill the skills gap and embrace the burgeoning digital economy”.

"The WIPO report has also provided a helping hand to developing countries by revealing the top players in AI from industry and academia that could be used for scientific cooperation as well as technology and best education practices transfer for their universities,” Alfrmawi said.