Experts suggest cultivating the service sector and other industries in advance to cope with the AI employment shock
At the 2026 Bund Summit, Cai Fang, a member of the Academic Divisions of the Chinese Academy of Social Sciences, proposed policies in four areas—promoting AI equality, guiding AI well, cultivating certain industries in advance, and resolving structural employment contradictions—to cope with the employment shock that AI may bring in the future. Data from the International Labour Organization show that the global unemployment rate is at a historic low, and in an OECD survey of 8,000 employers in 14 countries, only about 1% to 2% of companies adopting AI said they had reduced staff because of AI. Cai Fang suggested cultivating in advance the service sector, which has high demand elasticity and relatively low labor productivity, while Du Yang suggested that the tax system should favor companies that employ more people.
Body · AI translation
According to International Labour Organization data, the global unemployment rate is currently at a historically low level, but amid the technological transformation driven by artificial intelligence (AI), the employment shock remains a sword of Damocles. How should we respond to a future that is not yet clear? During the 2026 Bund Annual Conference, "preparing for a rainy day" was the common ground among experts, and on specific response measures, expanding the service sector and adjusting tax system arrangements were among the policy tools under discussion.
"Everyone is now talking about AI's impact on employment. But whether in China or other countries, so far no employment shock from AI is visible in the macro data," pointed out Cai Fang, a member of the Academic Divisions of the Chinese Academy of Social Sciences. However, at the micro level, recruitment data show that the number of new hires needed for programming positions has declined somewhat, and some employers have begun to wait and see—neither laying off workers nor hiring new ones... Therefore, the current employment situation is more like a "brief quiet period."
Mark Pearson, Acting Director of the Employment, Labour and Social Affairs Directorate of the Organisation for Economic Co-operation and Development, corroborated from survey data that "the employment doomsday is still far off": a survey of 8,000 employers in 14 countries showed that among employers that have already adopted AI, only 11% have cut jobs, and among employers that have cut positions, only 9% reduced employment because of AI. In other words, among all enterprises that have adopted AI, only about 1% to 2% say they have reduced staff because of AI.
The "J" curve was proposed by scholars such as Erik Brynjolfsson of Stanford University and mainly describes AI's impact on productivity: in the early stage, enterprises invest heavily to adapt to new technologies, which may even cause productivity to decline temporarily, and only after crossing a certain "critical point" does productivity begin to accelerate.
Cai Fang cited this research to explain the importance of "preparing for a rainy day." "What we see and what is happening today is not at all what tomorrow will bring; calculating today which positions and which jobs have high exposure is also only about today—we cannot predict what will happen in the future. In this sense, making policy preparations now and building a response platform at the institutional level is clearly more important," Cai Fang emphasized.
He believes that policy intervention should focus on common goals and common means, and in this regard he mentioned four aspects: promoting equal AI rights, guiding AI well, cultivating certain industries in advance, and focusing on resolving the current structural employment contradictions, so as to respond to the employment shock that AI may cause in the future.
Among them, the industries that can be "cultivated in advance" are those with high demand elasticity but currently seemingly low labor productivity, such as the service sector. Cai Fang believes that moderately decoupling industrial development, job creation, and people's employment and income from individual labor productivity will be the direction of the future and also the answer to the question of "where humanity is headed"; at the same time, this is also a buffer and firewall against employment shocks.
"New demand, new markets and new jobs all come from people's all-round development," Cai Fang said.
Regarding this view, Du Yang, director of the Institute of Population and Labor Economics at the Chinese Academy of Social Sciences, further explained that this means that in the AI era, the Baumol effect may no longer be a "Baumol's disease," but rather a way for people to share the productivity gains of AI.
She gave an example: employment in the film and television production industry fell from 1.3 million to 700,000 because many processes can be done with AI; but at the same time, new industries have emerged and are thriving—the short-drama industry has absorbed more than 1 million new jobs. Therefore, total employment in this industry has actually increased, and it has increased in an equal-rights way: it has not given all the jobs and income to the top film and television companies, but has allowed many young people to participate in producing short dramas watched by only about 10,000 or a few thousand people and still earn income.
She suggested that the country's overall tax system arrangements should favor enterprises that can employ more people. Because in the AI era, replacing people with equipment will happen automatically, and the government should encourage enterprises to hire more people through the tax system, which will be a very important institutional arrangement. "Our existing tax policies encourage enterprises to invest more, but mainly to invest in equipment—VAT rebates encourage buying more equipment, which is effective for encouraging production; but for enterprises that hire more people, there is no special institutional arrangement."