Fight for AI talent, a battle between China and US.
China and the U.S. are scrambling to secure one of the world’s most coveted resources: talent in artificial intelligence.
About 1 million AI practitioners are needed worldwide, but only 300,000 are currently working, according to a report by a research institute affiliated with Chinese internet giant Tencent Holdings. Although 370 educational institutions perform AI-related research globally, only about 20,000 graduates are produced in the field each year.
Companies are crossing borders and oceans to fill this 700,000-person gap. Google set up an AI center in Beijing this spring to recruit innovative students in China. Tencent and rival Alibaba Group Holding attended an AI conference in the American city of New Orleans in February to recruit students looking for a workplace to demonstrate their specialized knowledge.
China and the U.S. are competing furiously for the edge in AI technology. In a report released last summer, China vowed that its AI industry would be on par with developed nations in 2020 and a world leader in 2030. The U.S., for its part, held an AI summit in May with top engineers at the White House, which promised to preserve American leadership in the field. Current escalating trade tensions between the two countries could also heat up their fight to secure talent.
Companies are dishing out hefty salaries to secure top talent. Data scientists at Facebook make an average annual salary of around $400,000, said Tomoe Ishizumi, CEO of Palo Alto Insight, an AI service provider. Similar positions at Google and Amazon.com offer roughly the same pay. “Even large companies like IBM are having trouble securing talent,” she said.
Meanwhile, Japan has been unable to keep up. The average information technology employee makes just over $100,000 in the U.S. but only about half that in Japan, according to the Ministry of Economy, Trade and Industry. IT jobs pay about 140% better than all other industries in the U.S. but only 70% better in Japan. In China and India, that figures reaches seven to nine times more.
In an economy ministry survey of IT workers, about 7% of respondents said their company’s compensation system is based totally on seniority while more than half said a seniority system was in place with room for differences in ability and results. Japan will be unable to compete for global talent without a more flexible, meritocratic system — a reality some companies are beginning to tackle.
“We are looking for seven geniuses and 50 people of talent,” tweeted Yusaku Maezawa, president of Start Today, which operates online fashion marketplace Zozotown. He offered “geniuses” a maximum of 100 million yen ($908,000) a year.
In January, Toyota Motor promoted Gill Pratt, a former manager of AI-related projects at the U.S. Department of Defense, to a new executive position. The automaker also set up a company in March to develop cutting-edge technology for automated driving. It has already invested more than 300 billion yen and is introducing a new personnel system to attract AI talent.
Few Japanese companies, however, provide large budgets and attractive research opportunities. Government involvement will therefore be crucial if Japan wants to bring in more AI practitioners.
Shiga University in western Japan set up a data science department in April 2017. Similar departments newly created at Yokohama City University and Hiroshima University were also flooded with applications. Nevertheless, only about 2,800 people finish master’s-level programs in Japan each year.
“Japan could improve its strengthens in fields like robots with AI,” said Yutaka Matsuo, a specially appointed associate professor at the University of Tokyo, who stressed the need for government involvement in developing AI talent.
About 163 trillion gigabytes of data is expected to be created in 2025, a tenfold increase from 2016. With their knowledge of advanced mathematics, statistics and data processing, AI engineers will be the brains behind how to best utilize this massive amount of data as more businesses digitize.