Chinese experts have warned about the potentially significant impact that artificial intelligence (AI) could have on the labor market, calling for improved governance of the technology.
Currently, AI's efficiency doubles every seven months. As a result, models that could previously complete only one minute of coding work can now handle an hour's workload, lending credence to a conjecture by the "Godfather of AI" Geoffrey Hinton, who once called it the "single most impactful technology of our time," and expressed his concern about AI’s capability to "replace many, many jobs".
Facing these fast-paced advancements, human society must act quickly to prepare for the consequences of even more capable models, according to Gong Piming, an associate researcher at the Academy of Macroeconomic Research under China's National Development and Reform Commission (NDRC).
"Our current level of social adaptation is actually lagging behind the pace of AI development. The exponential acceleration of AI has become an indisputable fact. Such advancement is non-linear in nature, driving a leapfrog transformation in technological evolution," said Gong.
According to research by an American consulting firm, approximately 30 percent of global jobs will undergo significant transformation by 2030. Meanwhile, a report from a Singapore-based human resources research institute indicates that 53 percent of Singaporean companies plan to slow down or halt entry-level hiring over the next three years.
"Work that can be replaced by AI has certain characteristics. For example, tasks with very clear rules, heavy reliance on data, and outcomes that are readily verifiable. These tasks allow for straightforward verification of correctness, adherence to formatting standards, and consistency in internal logic -- all areas where AI excels at rapid inspection and correction," said Professor Zhang Ming from the School of Economics and Management, East China Normal University.
In the current climate, strengthening research on the protection of workers' rights has become a critical task.
The human resources authorities in Beijing recently released their top 10 most typical labor dispute arbitration cases in 2025. One of these cases involved a data collector who had worked at a technology company for 15 years and was responsible for traditional manual map data collection.
When the company transitioned to AI-driven automated data collection, eliminating the need for manual collection, his department and position were eliminated. The company terminated his employment contract, citing a significant change in the objective circumstances that formed the basis of the original employment agreement.
"The arbitration case is essentially a microcosm of the gap, or perhaps deviation, between technological advancement and societal adaptation. The core conflict lies in companies prioritizing efficiency by deploying AI as a tool to boost productivity, thereby favoring AI over human labor. Social equity, however, demands that the benefits of technological progress be shared by all citizens. This simplistic approach of screening or laying off workers demonstrates a failure to properly balance the requirements of social equity with corporate efficiency," said Gong.
The arbitration committee in this case determined that such changes initiated by the company lack the characteristics of force majeure and unforeseeability required by objective circumstances. The company directly shifted the operational risks arising from technological iteration onto the employee and terminated the employment contract on this basis, lacking factual and legal grounds. Therefore, the company was ordered to pay the employee compensation for unlawful termination of the employment contract.
"Companies actually have other choices. For instance, they could provide training to help individual employees adapt their skills to the transformation. Labor laws currently offer vague guidance on managing tech platform algorithms and defining responsibilities, potentially creating issues. However, these regulations can be promptly revised and refined in the future to keep pace with technological advancements and safeguard workers' rights," Gong said.
In August last year, China's State Council explicitly stated in an official document that artificial intelligence should be actively leveraged to create new jobs and empower traditional jobs, thereby mitigating its impact on employment.
Experts suggest establishing a risk warning system and labor protection mechanism to anticipate potential employment disruptions, ensuring proactive measures are taken before any adverse effects materialize.
"Our early warning concept aims to proactively address issues. For instance, when risks of job displacement due to AI begin to emerge, the government should provide employment guidance or resources. When AI substitution effects become pronounced, the government may need to prioritize training programs for workers in heavily impacted industries, alongside corresponding subsidies or assistance. Should an industry experience contraction, government measures may be activated to help affected workers navigate this transition more effectively," said Gong.
Furthermore, according to Zhang, a mechanism must be established to strike a balance between capital reaping the benefits of AI and workers bearing the costs of transformation, so as to ensure the inclusive sharing of technological dividends while clearly defining enterprises' social responsibilities during industrial transformation.
"In many countries, unemployment caused by AI is treated as a societal issue rather than an individual labor dispute, because individuals cannot stand against this historical tide. Businesses gain efficiency dividends from using AI, but we can't just hand all the gains to them - some should be transferred to employees. That's why academia is now exploring whether we can impose a robot tax, or AI tax on businesses with the proceeds specifically allocated to addressing unemployment risks caused by AI disruption," said Zhang.
Experts warn about AI's impact on employment, call for improved regulation
