Chinese companies and researchers now own two thirds of all machine learning patents, according to a new report from IP Australia, the government agency responsible for local filings.
Australia has filed just 0.16 per cent of the global total, according to the Machine Learning Innovation report, prepared in conjunction with industry body ACS.
IP Australia analysed machine learning patent filings from around the world, finding since 2012 36,740 patent families have been filed in machine learning technologies. 25,319 of them originate from China, with 96 per cent of those having no international involvement.
The next closest country, the United States, has filed 6,590 patents, around a quarter of China’s total. The US is followed by South Korea and Japan, with patient family filings then dropping off relatively quickly.
Australia is ranked seventeenth. Australian individuals and institutions have made just 59 patent filings, with most relying on international input, according to the report.
According to IP Australia, patents, which give exclusive commercial rights in exchange for public disclosure of new devices, substances or processes, are a good indicator of innovative activity and the maturity of an AI industry.
The report defines machine learning as “a form of AI that enables a system – or machine – to learn from data using algorithms”.
The vast majority of machine learning patent applications analysed by IP Australia are for image and video analysis, analytic and processing, and speech and text analysis. Applicants are commonly large corporations or universities and the leading industry for machine learning patents is telecommunications.
According to the report, machine learning patent filings have grown 27 per cent or more every year since 2012.
The AI gap
One reason for China’s dominance is a strong commitment from its government to fund and help commercialise AI and machine learning technology.
“The number of Chinese originating patents in machine learning may reflect Chinese government research and development initiatives,” the report states.
“These have been particularly strong in the field of AI, with a government venture capital investment of US $30 billion and state and public-private investments totalling tens of billions more.”
China has committed to creating a US$150 billion AI industry and to be the global AI leader by 2030.
Australia, meanwhile, has been relatively late to try and leverage the AI and machine learning boom and the federal government is facing criticism for failing to appropriately fund local development.
According to the latest AI roadmap from the government, digital technologies, including AI, would potentially be worth $315 billion to the Australian economy by 2028.
But the government has committed only AU$29 million over four years and last month rejected calls to increase AI funding, leading to Industry Minister Karen Andrews being labeled “all talk and no action” by her opposition counterpart, Ed Husic.
“The Morrison government’s roadmap talks up the importance of AI and then completely omits the key vehicle to get us there: a national AI investment plan,” Husic wrote in an AFR op-ed following the roadmap launch.
Minister Andrews launched the latest IP Australia report on machine learning filings this week.
“We know AI stands for artificial intelligence but this report shows it could easily stand for Australian Invention,” Andrews said in a statement.
“We’ve got an incredible history of ingenuity in this country and AI and machine learning are at the centre of that next generation of invention.”