The Chinese approach to closing the AI skills gap

Huawei, Yitu Tech, and Mai Mai reveal recruitment strategies for artificial intelligence talent
(Image: Stockfresh)

15 October 2018

China has rapidly risen to the status of global artificial intelligence (AI) superpower, but the businesses building the growth face the same struggles to find talent as their international rivals.

The Times Higher Education estimates that China produces more than twice as many research papers on artificial intelligence than any other nation, but the country drops to seventh in the number of AI technical professions, according to a report in Linkedin.

An analysis of AI talent in China by Mai Mai, the country’s own professional social networking site, revealed that China shares the common issues around the unequal distribution of talent.

Beijing is home to a whopping 60% of the talent pool, while the closest contenders of Hangzhou, Shanghai and Shenzhen each have only around 10%. No other Chinese city hits 2%.

The inequalities continue in the production pipeline. Harbin Institute of Technology, Beijing University of Posts and Telecommunications are the leading universities in AI talent training, but Tsinghua University and Peking University lag slightly behind.

The problem extends from emerging start-ups to established giants such as Shenzhen-based Huawei, China’s biggest manufacturer of mobile phones and telecommunications equipment.

Huawei rotating chairman Eric Xu believes AI talent distribution has taken the form of a Pareto Curve, as a small proportion of staff are creating the main business value, leading enterprise to rely on the innovations created by outstanding talent in order to grow. This means that people who can make breakthroughs are highly sought after.

Attracting talent
Huawei has an annual R&D budget of $15-$20 billion (€13 – €17.3 billion) that is used to develop better products and drive internal efficiencies. It also helps attract high-end talent searching for meaning in their work.

Huawei founder Ren Zhengfei outlined the approach at a progress briefing on the AI application in the company’s Global Technical Services (GTS) unit.

“Scientists research basic theories with an eye to the future,” he said. “They don’t have to consider the commercial use of their research results, but need to spread thought and theories and make breakthroughs for other domains.”

This desire for jobs that offer more than just money is growing. Huawei’s 2017 employee survey found that staff born in the nineties believe that self progression is the most important aspect of their jobs, followed by the meaning in their work, recognition and compensation.

To cope with these changes, Huawei has ramped up R&D spending and redefined its vision to reflect these shifting values. The vision now reads: “Bring digital to every person, home and organisation for a fully connected intelligent world.”

Top talent desires top salaries, but other workers can be replaced by technology. Huang Weiwei, a senior management consultant at Huawei and professor at the school of business of Reninmin University, says that Huawei has developed smart manufacturing that increased efficacy by 30% without hiring more staff. The money that is saved can then be spent on the most valuable staff.

The proportion of Huawei employees with a master’s and PhD degree at the company has grown by 10%, but Yitu Tech, an emerging AI giant in China, has found that qualifications do not always reflect an employee’s value.

“Ninety percent of our researchers are fresh graduates from bachelor’s degrees,” said the company’s chief innovation officer Lu Hao.

“We choose not to hire those PhDs, not by decision but [because] they failed the interviews miserably. I think that shows some of the actual facts of current education systems… Simply training a model is not something that difficult. With all the tools that are available, training a model is not super-hard science — even undergrads can do it.

“What they’re missing is attention to data, how [to] really understand and analyse data, and using intuition to do a lot of those things. Those are some of the things that are really missing when we interview a lot of PhDs.”

Mai Mai cofounder Qian Wang added that this could become common as the rate of development increases. She learned about facial recognition at university through studies in the classroom, experiments in the lab, and competitions abroad, but felt that the technology was still a long way from the mainstream. Just over 10 years later, it is becoming commonplace.

“In China, the greatest concern for me is education,” she said. “China doesn’t lack talent. The key is whether our education will keep up with the needs of AI.”

Role of the private sector
Universities also face financial barriers to developing skills in AI due to the cost of the technology. Yitu Tech could afford to spend 1 million renminbi (€124,801) to label the data in a machine learning model that they then gave away for nothing to the research community, but such costs are prohibitive for universities without sponsorship from the private sector.

Qian believes that companies should work together with universities to develop the talent that they may one day employ.

“Companies need to have some talent to work on AI research and the universities need capital and funding support and they also need to look for applications in [business] areas so that they can apply these scientific research results into real business scenarios,” she said.

Huang feels that employees need to be accountable for their own training rather than relying solely on their employer. He argues that universities are chiefly responsible for helping them to learn how to learn: “To truly transform themselves, individuals need to rely on themselves, so the ability to learn is very critical. Universities need to help the students to improve their learning capabilities so they can better transform themselves when it is needed.”

Qian adds that the skills they need go beyond academic abilities.

“If they are just bookworms or nerds who cannot adapt according to the times and cannot learn to apply what they know then they may find themselves out of a job,” she said. “And also they could have learnt and acquired many high-level technical skills, but maybe their EQ [emotional quotient] is very poor and they may not be able to keep up with the requirements of a new era, so they may not become the core expert or talent either.

“If we have a very good capability to learn and at the same time we’re able to keep in step with the changes in times, then we would be able to create value for our customers and our customers’ customers, and also for our companies.”


IDG News Service

Read More:

Comments are closed.

Back to Top ↑