AI schools blossom in China

As country strives to fill talent gaps, experts express concern about overhype in industry

All of a sudden, schools and institutions dedicated to artifcial intelligence (AI) teaching and research have blossomed in China, as the country strives to become a global incubator for AI talent by 2030. Official data shows that more than 70 Chinese universities and colleges had introduced AI-related majors as of December 2017, and that has helped to quickly bring the advanced technology into reality. However, education experts and insiders are now concerned that the rush will likely generate too much industry hype.

In the eyes of Xiao Jun, vice dean of the School of Artificial Intelligence under the University of the Chinese Academy of Sciences (UCAS), the establishment of AI schools is a “must.”

After more than 60 years of development, AI technologies – including cloud computing, big data analysis and data collection – have become so powerful and smart that UCAS thinks it is necessary to set up schools specifically for the study of AI technologies as well as to further integrate them in accordance with social and industrial demands, Xiao told the Global Times.

UCAS is the first university in China to set up an AI school. Consisting of six teaching and research offices that run a total of 75 classes, the school addresses areas including pattern recognition, intelligent human-machine interaction and AI fundamentals.

In September, 280 postgraduates, both full-time and part-time, will be enrolled into the UCAS School of Artificial Intelligence.

But the quota is not enough, and the school has asked for more, said Xiao.

Meanwhile, the School of Artificial Intelligence at Xidian University, founded in November 2017, also feels it has to scale up its enrollment numbers.

Because there were so many applicants, Xidian’s AI school ended up enrolling 249 students in the end, more than the original enrollment cap of 230, Wang Shuang, a professor with the school, told the Global Times.

Aiming high

Promising future career prospects have fueled students’ passion for AI schooling.

An AI engineer’s annual salary can range between 300,000 yuan ($43,925) and 600,000 yuan, with some even earning 1 million yuan if they have more skills or a solid educational background, according to a survey conducted by news portal tech.qq.com in March 2017.

Behind those high salaries is the fact that China is short of specialists to help drive the country’s booming AI market, so high salaries work as an incentive to attract talent.

China’s AI market was worth 23.74 billion yuan in 2017, up 67 percent year-on-year, and is estimated to grow 75 percent year-on-year this year, a report issued by Tsinghua University showed on July 13.

The domestic AI industry in China has attracted the most funding in the world compared to other countries and regions, accounting for 60 percent of all global investment in this sector from 2013 to the first quarter of 2018, but it still lags behind the US when it comes to the amount of AI specialists.

By the end of 2017, China had 18,232 AI specialists, accounting for 8.9 percent of the world’s total, while the US held a share of 13.9 percent, according to the report.

To fill the gap, China needs to double the current scale of AI talent – nurturing, an industry insider told the Global Times on condition of anonymity.

The past month has witnessed five AI schools be established across the country, with even more set to open in the near future.

One of the central government’s most recent initiatives includes the Ministry of Education’s release of an AI talent training plan in April.

According to the plan, China will strive to set up 100 university programs that combine AI and other subjects by 2020 and make national AI schools the core of global AI innovation that is capable of supplying China with technical support and experts.

China looks set to make breakthroughs and innovations in basic AI theories, which rely on AI experts who hold master’s or PhD degrees, according to the insider, noting that those who solely hold bachelor’s degrees can only do easy jobs like AI application development and can hardly take on tasks like algorithm research.

Reality behind the hype

However, Liu Ying, founder of Beijing-based online IT training platform Tuling, is not a supporter of the mass establishment of AI schools around the country.

“This ‘mass production’ mode is unlikely going to nurture AI talent, as it is hard to find so many qualified students for AI study,” Liu told the Global Times, noting that students who want to do research on AI must be proficient in both math and algorithms.

Furthermore, Liu and the abovementioned insider both think there is overhype in domestic AI school development, questioning whether all AI schools in China can be equipped with strong faculties and whether those schools can truly pipeline qualified talent to enterprises.

The insider noted that currently, some universities just want to jump on the bandwagon – opening an AI school to help enrollment – but in reality, their schools’ training schemes still focus on traditional computer science.

However, he remarked that setting up some AI schools nationwide can help expose hidden but fixable problems, which then benefits the overall development of the AI industry.

As AI is a cross-discipline subject, some universities build their AI schools by absorbing teaching resources from various related majors.

For instance, Xidian’s AI school originated from the university’s neural network research center, which was founded in 1990, with the researchers of the center becoming the foundation of the AI school’s faculty. The university also launched its Intelligence Science and Technology major for undergraduates in 2003, which has been providing students with experience and rich resources for the establishment of the AI school.

“Different from other AI schools which set up frameworks first and gradually build up teacher teams, our school was originally developed from a research team,” said Wang.

However, Xidian’s AI school plans to more than double the current teacher amount to 150 by recruiting more talent.

In the view of Xiao, ordinary AI schools are likely to face insufficient faculty experts for curriculum building, because AI is a professional, cross-discipline and application-oriented subject.

Seldom AI schools in China can have as powerful teacher resources as those of UCAS, which is backed by the Chinese Academy of Sciences, Xiao remarked.

How to better nurture AI experts is a key task faced by all AI schools, according to Wang. “Even though we have fairly mature training programs and our curriculum design has been consistently revamped with the help of major AI enterprises like Google and Baidu, we still cannot be sure that our graduates are perfectly ready for the market,” said Wang.

“AI involves many subjects and develops very quickly, thereby researchers or engineers in this sector must have a good habit of self-study and

maintain lifelong learning.”