China and the United States are ahead of the global competition to dominate artificial intelligence (AI), according to a study by the U.N. World Intellectual Property Organization (WIPO) published on Thursday.

The study found U.S. tech giant IBM had by far the biggest AI patent portfolio, with 8,920 patents, ahead of Microsoft with 5,930 and a group of mainly Japanese tech conglomerates.

China accounted for 17 of the top 20 academic institutions involved in patenting AI and was particularly strong in the fast growing area of "deep learning" - a machine-learning technique that includes speech recognition systems.

"The U.S. and China obviously have stolen a lead. They're out in front in this area, in terms of numbers of applications, and in scientific publications," WIPO Director-General Francis Gurry told a news conference.

U.S. President Donald Trump has accused China of stealing American innovations and technology and has slapped trade tariffs on $234 billion of Chinese goods to punish Beijing.

China said in December it resolutely opposed "slanderous" accusations from the United States and other allies criticizing China for economic espionage and stealing intellectual property and company secrets.

Gurry acknowledged there were accusations about China's behavior but there was no doubt it had embraced the global intellectual property system, with the world's largest patent office and the largest number of domestic patent applications.

"They are serious players in the field of intellectual property," he said.

The WIPO study analyzed international patent filings, scientific publications, litigation filings and acquisition activity, and found there had been as many patent applications for AI since 2013 as in the half century since the term was coined in the 1950s.

Patent applications in machine learning, which includes techniques used by ride-sharing services to minimize detours, averaged annual growth of 28 percent between 2013 and 2016, the last year for which data is available, because of an 18-month period before confidential applications are publicly disclosed.

Much of that growth came from deep learning, which overtook robotics as it ballooned from 118 patent applications in 2013 to 2,399 in 2016.

The single most popular AI application was computer vision, used in self-driving cars, and mentioned in 49 percent of all AI-related patents.

The study showed how technology had followed science, Gurry said, with the 2013 boom in technological applications coming 10 years after a similar surge in scientific publications.

However, the world did not have any reliable way of measuring the quality of patent applications.

"If you did, you wouldn't need a venture capital industry," he said.