Amid growth in AI courses, concerns over a skills mismatch

Despite huge government and private funding for expansion in new-generation artificial intelligence and a wave of new AI departments and courses being established at Chinese universities, some experts fear China may soon have an oversupply of AI graduates due to a mismatch between what is taught in universities and the requirements of the high-tech jobs market.

China’s universities have rushed to heed the call for AI courses, setting up a wave of new AI schools and relevant majors since 2017.

According to incomplete statistics, 43 AI schools and research institutes can be found across Chinese universities today, while the number of universities offering undergraduate majors in AI exceeds 500.

In late August, the Ministry of Education announced speeded up efforts to incorporate AI into the national curriculum in high schools. Spurred by the government’s emphasis on AI, the aspiration to study AI has grown among Chinese students who sit the Gaokao university entrance exam each year.

However, like many other industries, the tech sector has this year faced rounds of layoffs as China’s big tech companies report plans for further downsizing to cut costs. In many cases, entry-level hires are first in the firing line, adversely affecting the prospects of those just starting out.

Buyer’s market

Recently, Chinese media have also begun to report on the predicament of young workers, who say the landscape of AI jobs is already a buyer's market.

Qin Zengchang, a professor of AI at Beihang University, says this rings true even for his postgraduate students at Beihang, China’s preeminent institution for science and technology.

“Students are getting fewer job offers, although employment is, in principle, not an issue for this industry,” he told University World News.

However, he cautioned against putting all the blame on universities. The broader “business and societal culture” which favours those with existing experience is not helping fresh graduates’ chances, he said.

But others point to an alarming mismatch emerging between the skills of new graduates as universities struggle to keep up with new developments in AI and lack resources for research that incurs huge data costs and for state-of-the-art equipment.

The speed at which AI courses are being set up, and a shortage of experts to teach them, has led to experts criticising some universities for offering courses which may be sub-par.

A predicted shortage

Until recently, official media painted a picture of an alarming shortage of AI talent. Various studies predicted a sizeable shortfall of four to five million specialist graduates required to meet the needs of a strategic industry expected to grow by leaps and bounds.

Not just the central government in Beijing but also provincial and local governments have recently rolled out generous subsidies to AI enterprises, for example, for rising data centre costs, as part of the country’s AI growth strategy.

In Beijing, home to almost 40% of the country’s AI companies, the municipal government provided ‘computing vouchers’ worth RMB57.5248 million (about US$8 million) to 60 companies in the first half of 2024.

The vouchers can be used to buy time at AI data centres to train and run large language models (LLMs) used in generative AI.

Last year, a report by US management consultants McKinsey and Company estimated demand in China would require an extra 6 million skilled AI professionals by 2030 — a five-fold increase from 2022.

This projection points to a staggering shortfall: the talent pool of graduates with relevant qualifications emerging from domestic universities was estimated at just 2 million.

Difficulties recruiting suitable talent

Yet, McKinsey’s survey of AI business executives from 102 Chinese companies found more than half of respondents had difficulties finding suitable talent to fill AI-related job vacancies.

According to the report, roles such as data engineers, data architects, and machine learning engineers are the most critical need.

The Chinese government increasingly views talent as vital for technological advancement. President Xi Jinping has repeatedly referred to talent as “the first resource” in China’s quest for self-sufficient innovation.

The government’s multi-pronged talent plan includes enhancing domestic education, luring back Chinese talent from overseas, and attracting international skills.

During a national science conference in June, Xi emphasised the importance of filling the gaps. “Some key core technologies are controlled by others, and there is a shortage of top scientific and technological talents,” he acknowledged.

According to multiple sources who spoke anonymously to University World News, universities are constantly struggling for resources. This includes “computing power, data, and models”, a young engineer in computing technology, who prefers not to be named, said.

Shortages at universities are often the result of funding constraints, as the high costs of GPUs (graphic processing units) due to export restrictions imposed by the United States, are a major deterrent to upgrading equipment and facilities.

In 2018, the Ministry of Education rolled out the “Innovation Action Plan for Artificial Intelligence in Higher Education Institutions”, which outlined the content of the AI major in China.

But in a sign that they understand they need to produce more employable graduates, universities are taking steps to improve the AI curriculum by strengthening industry linkages to reflect actual market demand in training and course content.

Industry exposure also helps fill gaps in areas where universities are behind the curve.

High paying jobs but stiff competition

Nonetheless, AI-related jobs are among the most well-paid across industries.

On 3 July, recruitment platform Zhilian Zhaopin released the “China Enterprise Recruitment Salary Report”, which found that in the second quarter of 2024, AI led all industries in starting salaries, with a monthly average of RMB13,594 (about US$1,910).

That starting salary represents a 5.3% year-on-year increase compared to the all-industry average of RMB10,313 (about US$1,449).

Among all AI-related roles, AI engineers in algorithms, C language, embedded software development, and machine vision receive the highest pay, with an average monthly salary of RMB22,003 (US$3,090).

The report concludes that those employed in AI have strong high-tech capabilities, a high level of education, and high salaries.

Algorithm engineers, software developers and machine learning engineers are most in demand, but the barriers to entry are high.

“Graduates from non-elite universities are often seen to be incapable of the tasks, or they need to have a masters degree at least,” said Wang Mengmeng, who until recently worked at iFLYTEK, a leading Chinese AI company.

After graduating with bachelor and masters degrees in mathematics, Wang initially interviewed for an algorithm engineer position with iFLYTEK, which originated at the University of Science and Technology of China.

But he was offered a role as a software product manager in its EdTech department, where he was responsible for the R&D, design, and operations of relevant AI software.

As with many other tech giants, the “996 culture”, which puts employees under pressure to work overtime (9am to 9pm six days a week) prevails at the company.

iFLYTEK, known for its voice recognition software, had vowed to catch up with OpenAI’s AI model “within a matter of months”. On 28 June, it launched the upgraded version of the “Spark 4.0 LLM”, which it says is on par with GPT-4 Turbo, OpenAI's current top model.

High costs of success

However, the achievement may come at a cost. Less than two weeks before the announcement, news broke of the sudden death of an employee – a senior engineer passed away at the age of 38 after suffering a heart attack.

For Wang, the event touched a raw nerve; in shock, the 29-year-old decided to quit.

On social media platform Weibo, the employee’s family has alleged that the death is related to overwork and filed for compensation from iFLYTEK.

According to media reports, an insider told Southern Metropolis Daily reporters that iFLYTEK “has handed the matter over to the police”.

In an internal notice to staff, it said the employee had died after falling ill at home and that the company has set up a working group to apply for union assistance funds, contact and handle commercial insurance, assist the employee's family in “handling related follow-up matters and help the family overcome this difficult time”, according to one report.

Now back in the job market, Wang described employment prospects as “average” given job cuts amid China’s continuing economic downturn. “Over the past two to three years, there have been fewer openings and higher standards from employers,” she told University World News.

Looking ahead, Wang says the emergence of OpenAI or iFLYTEK’s Spark may in fact become a driver of the industry's own redundancies.

“For example, before, you may have needed 10 front-end or back-end personnel to write the code. Now, with the help of OpenAI, you might only need eight, so a couple of people can be cut, and manpower can be reduced,” she said.

There is already some evidence highlighted in official media that the enormous demand for a high-tech workforce is putting jobs in traditional industries at risk, a dual effect that the academic community calls a “structural dilemma” that will require continuous training both at universities and outside them.