Shanghai Observer Commentary: Behind the Rumors of Lao Gan Ma's AI Quality Inspection Layoffs, How to Resolve AI Employment Anxiety

With the launch of the AI visual quality inspection system, only 48 of the 600 pepper-sorting workers remain on the assembly line—Lao Gan Ma surely never imagined that what was meant to be a positive publicity story about technology empowering enterprise development would end up pushing the company into the center of a public opinion storm. Although the company responded that the claim of "laying off more than 500 workers after using AI" is false, public doubts have not dissipated, and Lao Gan Ma's online sales have also seen a marked decline. Whether the layoff rumors are true and how to restore consumer trust both await further confirmation and action. But the reason this incident occurred must be seen in the context of the increasingly intense AI anxiety behind public opinion. The impact of AI job displacement has reached a point where it must be confronted directly. 1 In his 1845 book The Condition of the Working Class in England, Engels described the phenomenon of simultaneous leaps in productivity and worker poverty during the Industrial Revolution. This phenomenon has been called the "Engels' pause" by economic historians and has prompted reflection on how technological change can better promote social fairness and justice. Times have changed, but the old problem still exists in a new context. According to data from the global tech layoff tracking database Layoffs.fyi, in the first half of 2026 alone, the cumulative number of layoffs in the global tech industry almost matched the level of all of 2025: hundreds of layoff events have occurred in the global tech industry, involving 202 tech companies and more than 120,000 cumulative layoffs. Of course, some analyses also point out that some layoffs are likely companies "blaming AI" for original business adjustments, and the real reasons go beyond this, but this is enough to show that expectations and anxiety about AI coexist in the industry. In traditional industries such as manufacturing, some repetitive processes are also being taken over as embodied intelligent robots enter workshops. Basic operations, financial accounting, and telemarketing are also areas where recruitment demand has shrunk. Some will retort, "It's not AI replacing people, but people who can use AI replacing people who can't." Others believe that while AI impacts traditional jobs, it will also create more and higher-quality jobs, just as previous technological revolutions and industrial transformations did. These views certainly have merit, but the actual situation is often more complex. For example, the idea of "destruction before creation" treats AI job displacement as an idealized process of replacing the old with the new, but in fact, the destruction of traditional employment by new technologies usually comes quickly and in concentrated form, while the creation of new jobs is often delayed and dispersed. Those affected are one group of people, and those who benefit are another; the two cannot simply offset each other. Another example is that AI substitution is not necessarily reflected only in the number of jobs, but may also be reflected in employment quality, including lower wages, degraded working conditions, and a large amount of dispatched employment and unstable employment. 2 So, does that mean the situation is already very severe? Is that so? It is autumn recruitment season again. During this period, Shanghai has held many offline job fairs one after another. In the bustling crowds, I felt an employment atmosphere somewhat different from the one dramatized and exaggerated by self-media. Fresh graduates are of course somewhat confused and anxious. A science and engineering student holding two patents and ranking in the top 3% of his major by GPA said bluntly, "I have a sense of AI crisis." What he worries about is that his AI level is insufficient and that he will be "squeezed out" by people who are better at using AI. From the demand side, more and more employers are writing AI application ability into job descriptions, and requirements for job seekers' AI literacy have extended from technical positions to non-technical positions. But the core of the job market is always "people," not AI. "Everyone says there are so many graduates now, so why can't we recruit suitable people?" Some HR staff complain that alongside "difficult employment" is "difficult recruitment." In other words, the difficulty is not that there are no people or no jobs, but the matching of people to jobs, that is, the structural employment contradiction often mentioned by experts. In theoretical deduction, a job can be broken down into multiple capability dimensions and matched one by one with the capabilities mastered by AI, thereby concluding whether the job is highly exposed to AI. This is how the widely circulated "list of occupations eliminated by AI" came about, and many jobs have been labeled "high-risk," triggering widespread employment anxiety. If "matching" in reality were really that simple, and if AI could easily eliminate most people, companies would not be so troubled by it. After all, judgment ability, emotional perception ability, aesthetic ability, and other qualities remain unique advantages of humans, and they are also qualities that are less explicit in job requirements but crucial in actual work. At job fairs, job seekers and companies communicate face to face, each with their own expectations and requirements, reaching agreement through adjustment. Real employment happens in this kind of back-and-forth—and as things stand, this is still the mainstream. 3 We should not exaggerate the impact of AI job displacement, nor should we evade the contradictions. What matters now is what should be done. Some economists point out that we are currently in a quiet period before AI produces a greater employment impact, and it is a window period for analyzing AI's employment impact and preparing theoretical and policy responses in advance. To respond to the impact of AI on employment, we clearly cannot count on technology to correct itself; government intervention is needed, with all parties participating together, to promote AI for good. Just two days ago, the General Office of the State Council issued the Opinions on Adapting to the Development of AI Technology and Helping Build an Employment-Friendly Development Model, emphasizing the need to actively enhance the positive effects of AI in creating employment opportunities, empowering traditional jobs, improving employment quality, and stimulating entrepreneurial vitality, effectively mitigating the employment substitution impact of AI applications, and striving to achieve inclusive development in which technological progress and industrial upgrading are coordinated with employment promotion and livelihood improvement. Accelerating the development of new occupations in the AI field, establishing an employment risk assessment mechanism for AI applications, improving workers' digital literacy and employability, and improving the systems for protecting workers' rights and interests and social security... These measures are not meant to put the brakes on technological development, but to try to establish a buffering and assessment mechanism between technological progress and employment promotion, repositioning "people" from a passive cost item to the subject of transformation. Around us, some changes have already taken place. For example, how to systematically improve AI capabilities is a common challenge faced by many people. This year Shanghai launched training and certification for AI promoters. Relying on a systematic training and certification system, it popularizes AI application skills among practitioners in all walks of life and improves the public's digital literacy and practical abilities. By the end of June, training and certification had been completed for more than 50,000 person-times of AI promoters. Next, the government and universities will provide general AI capability training to improve the market matching degree and allocation efficiency of human resources; enterprises will undertake more specialized AI skills training that better fits industry characteristics, guiding AI to enhance rather than replace workers' abilities—this will become a feasible path forward. After all, only by fully giving play to human subjectivity can we, amid the high uncertainty brought by AI, clarify why we develop and for whom we progress, so that technological development truly benefits the broadest masses of workers.