Asian startups, including China-backed ventures, are struggling to keep pace with Western rivals in several spheres of artificial intelligence (AI) development, new research shows.
And their reluctance to embrace and endorse emerging industry standards may eventually leave them isolated from potential strategic partners and blocked from particular market segments.
China leads the pack in Asia with 383 AI startups, according to studies by consultancy Roland Berger and investment firm Asgard Capital, both based in Germany. However, Beijing is still well behind the US, home to 1,393 startups, according to the same research.
Japan (113 startups), India (82), South Korea (42) and Singapore (35) are the other Asian pacesetters; Australia has 27 startups in Oceania. Israel (362), the UK (245) and Canada (131) are also in the top five worldwide.
Beijing, Tokyo, Shanghai, Shenzhen, Seoul, Singapore and Bangalore, which collectively account for 501 AI startups, are competing for the status of regional AI hubs, though China’s aggressive state backing means it will probably be no contest.
China’s government has invested US$2.3 billion in an AI research center in Beijing and a further US$5 billion in the nearby port city of Tianjin.
Accounting by some estimates for more than half of all global AI investment in the last five years, China hopes to match US technological levels in the industry by 2020. China declared in its ‘2017 Next Generation Artificial Intelligence Plan’ that it aims to be the world’s innovation leader by 2030.
To be sure, Chinese digital conglomerates Alibaba, Tencent and Baidu are starting to match the global presence of Western tech giants like Facebook, Google, Apple and Amazon.
“These [Chinese] companies have all the necessary ingredients: access to data, the financial means to invest in the most up-to-date technology and buy data, sophisticated search engines, control over the IT infrastructure and the ability to attract a skilled labor force,” the German researchers noted about China’s tech titans.
They are also dragging along promising new startups like SenseTime (China), Appier (Taiwan), AIDA (Singapore), AdMov (the Philippines), Roadstar.ai (China), Datanest (Indonesia), ViSenze (Singapore), and Sigtuple (India).
But the failure of Asian AI institutions and companies to support global efforts to create a regulatory framework for the emerging industry means they could eventually be governed by a system of ethics and standards shaped by Western ideals; there will be little opportunity to opt out if they want to work with global partners.
Sony Corporation is the only Asian company involved in the so-called ‘Partnership on AI’, a collaborative forum between academics, industry leaders and civil organizations that aims to ensure AI research is transparent and accountable, is beneficial to the world economy and has social safeguards.
University of Tokyo and the Hong Kong University of Technology are Asia’s only representatives from academia, while the Center for Internet and Society (India) and Digital Asia Hub (Hong Kong) are the only collaborating non-profit groups.
Asian legal experts made a limited contribution to a landmark set of global standards released in 2017 by the US-based Institute of Electrical and Electronics Engineers, though they still have the chance to take part in ongoing discussions on specific industry sectors influenced by AI.
Some countries are developing their own sets of rules, though this may ultimately detract from efforts to get a global consensus. Singapore said in June it would form a panel to advise the government on AI ethical issues, but it will still have the flexibility to incorporate overseas frameworks.
One obvious reason Asians may be holding back is the blurry line between AI systems’ commercial and security applications, which has thrown up a raft of ethical issues over data-sharing and privacy.
SenseTime, now the most valuable private AI start-up with a valuation of more than US$3 billion, made its name developing bank card verification systems. But its facial recognition software is also being used as part of a huge surveillance network that allows China to better track and monitor ordinary Chinese citizens.
The ability of Asian companies to go it alone on AI is questionable, as even top-spending China lags significantly in many systems Deciphering China’s AI Dream, a study by University of Oxford researcher Jeffrey Ding, gave China a score of just 17 out of 100 for its share of the world’s AI capabilities, compared with 33 for the US.
China leads the US in data collection, but is behind in the other three key pillars: hardware, research and the commercial sector, the Oxford study found.
Chinese companies are trying to buy into American and European firms, but the doors are closing fast as regulators recognize the threat they pose, at both commercial and security levels. It is no accident that technology shipments are at the heart of the simmering US-China trade showdown.
Ding admitted that there was one field of market strength the Chinese firms could harness when all else failed: “Industrial espionage and state-directed IP theft are definitely major issues,” he said.