Liu Jinsong, Senior Expert, Tencent Research Institute
While white-collar professionals are still anxious about whether their skills will be distilled by companies, over two-thirds of the short drama actors waiting for roles in Hengdian are facing a situation with no scripts to perform.If three years ago, the anxiety over AI job displacement was a "cry wolf" prophecy, laced with a sense of thrill and observation amidst the danger, now that displacement is concretely happening in a growing number of fields.In basic programming, 75% of new code is generated by AI; in customer service, 80% of routine inquiries are handled by AI; in translation, 90% of the demand has vanished, with human translators now undertaking the final proofreading tasks.If white-collar positions like programmers and translators have room to maneuver to other fields after being replaced by AI—as is often joked about, delivering food or driving for ride-hailing services after unemployment—what happens when even these fundamental, safety-net jobs are taken over by AI? How should we cope then?As autonomous delivery vehicles and self-driving cars cross the threshold of widespread application, job displacement in these two sectors might also transform overnight, just like in the short drama field.As a potential exploration to address AI's employment impact, alongside the idealistic Universal Basic Income (UBI), perhaps it is time we discuss the "Human Quota System."Starting with Extras "Selling Their Faces"
Recently, two trends in the film and television field are particularly noteworthy. One is the news about content platforms establishing "AI Talent Libraries," frequently making headlines; the other is about short drama actors losing their jobs, occasionally mentioned by media, with many former "dominant CEO" types reportedly returning to their hometowns to farm.Both, from different angles, bear witness to the transformation the film and TV industry is facing under the impact of AI. The original intent of creating "AI Talent Libraries" is to reduce shooting costs and free up actors' time and schedules. But for the actors, once digitized by AI, it brings not only an increase in the volume of works but also a potential dilution of their value.Thus, on one side, platforms are aggressively promoting AI adoption, while on the other, actors are denying involvement and opposing it. Many netizens also believe that compared to electronic tears and AI-generated people, they prefer watching real human performances. Although AI can precisely simulate the trajectory of tears, the curve of a smile, and even analyze big data to produce the "emotional template" most likely to move audiences, it cannot yet simulate human authenticity and soul.After all, a great performance often involves an actor's unique understanding of the character, projections of life experiences, and even sparks of improvisation.Backed by their influence and fan support, A-list stars naturally have the leverage to refuse when facing the AI wave. However, for short drama actors with less market influence, especially extras at the bottom of the industry chain, there is almost no capacity to resist or room for negotiation when confronted by AI's incursion.Recently, in Hengdian's extra circles, the continuously falling price of "selling faces" has made everyone tangibly feel the chill brought by AI replacement. "Selling faces" is just a joking, self-deprecating term. Essentially, similar to a "Celebrity AI Copyright Library," it involves granting an extra's portrait rights to a production company via AI methods, with the company paying a certain fee for the portrait rights usage.According to media reports, production companies initially priced "buying faces" typically between 1,500 to 3,000 RMB, but it has recently dropped to 100 RMB per year. The low-cost "sale of faces" is clearly not a mutually agreed-upon fair deal, but a helpless act by the most grassroots practitioners amid a sharp decline in live-action short drama filming.The traditional extra market, despite having a low entry barrier, required you to arrive at the set on time, follow the director's instructions, cooperate with the blocking, and sometimes endure rain, scorching sun, or standing all day. This hardship itself constituted a form of physical scarcity and irreplaceability, which also determined the price floor for human extras.And the AI-mediated "face-selling" model precisely shatters this underlying logic: one scan can generate an infinite number of "you." In a time-travel drama, you are Bystander A; in an urban drama, you are Background Figurine B; in a suspense drama, you are Zombie C. Yet, none of this has anything to do with the real you.Once an extra is abstracted into "data," it inevitably follows the logic of data pricing, with marginal costs approaching zero and supply nearing infinity. The only trajectory for the price is continuously downwards.From the production company's perspective, this indeed achieves cost reduction and efficiency gains. However, from a societal viewpoint, AI's involvement signifies a complete reversal of the supply-demand relationship in the extra market. In the torrent of AI, an extra's face is no longer a performance resource that must be physically present, but disposable data material that can be arbitrarily captured, replicated, without the need for subsequent royalty payments.The plight of extras is by no means unique to the film and television industry. It could be a metaphor for many industries under the future impact of AI: once your labor can be digitally replicated by AI, pricing power no longer belongs to the laborer, but to computing power, and the price of computing power is far lower than that of the vast majority of human labor.Extras, customer service agents, translators, designers, and basic programmers will be among the first group of practitioners hit by AI, but they will not be the last.The Proposal of the "Human Quota System"
If more industries experience concentrated job displacement under the impact of AI, like the short drama field, how should we respond? Historically, every technological revolution destroys old jobs but also creates new ones. However, past technological changes often involved a long evolutionary process. The steam engine revolution took nearly a hundred years from its beginning to full societal absorption; the electrical revolution took half a century.But the speed of AI replacement is much faster; once it begins, it might only take months. Taking the short drama content field as an example, before the 2026 Spring Festival, many production crews were still ambitious, planning their filming schedules for the new year. But after the festival, with the maturation of video generation technology, the once-booming live-action short drama industry seemed to be instantly decimated.Many production companies either suspended live-action filming plans or directly transitioned to AI content creation. Looking at content supply efficiency, the monthly volume of AI-produced content now online has surpassed the scale of live-action short dramas from the previous year, achieving more than a 10-fold increase in production efficiency.Of course, the emergence of AI-generated content also brings new job opportunities, such as prompt engineers and card drawers. Even existing roles like screenwriters, directors, and editors can quickly pivot and transition seamlessly. However, between the shift from old to new occupations, significant discrepancies often exist regarding demographics, region, and timing.In the AI short drama production chain, investors, directors, and post-production editors are still needed. Their transition from live-action drama to AI short drama production is merely a change of track; the process changes, but the narrative logic remains unchanged. But for many other practitioners in the industry chain, like extras, photographers, lighting technicians, art directors, makeup and costume designers, and set runners, they are not so fortunate.Therefore, for different groups of people, the transition from old to new jobs brought by AI is like an invisible watershed of destiny: some will seize the window of technological dividends and gain a premium from skill revaluation; but others will face a life test requiring them to start over from scratch. Just like extras "selling their faces"—a one-time buyout, usable for a year or a lifetime. Your face might continue working, but you personally are unemployed.And this is only the most visible form of displacement. A broader impact is happening in places we do not see: back-office auditing in banks, junior positions in law firms, customer service teams in e-commerce companies... This is a vast, silent, and often overlooked group that not only constitutes a significant foundation supporting the stable operation of society in real life but also forms the most stable buffer layer in the labor market.When this buffer layer begins to be impacted, or even breached, AI's effect on employment is no longer just a temporary pain for a specific industry but evolves into a societal proposition.Facing this potential crisis, OpenAI founder Sam Altman advocates for implementing Universal Basic Income (UBI) to secure basic living needs. Elon Musk suggests directly issuing checks to achieve universal high income. Their advice might apply in certain scenarios but may not suit all national conditions.A more critical question is: handing out money solves the problem of survival, but not the problems of human existence, subjectivity, and sociality. Therefore, rather than just distributing money, we perhaps need to think more about another matter: it is time to discuss the "Human Quota" system.Its core logic is to require, through legal provisions, that industries retain a certain percentage of human jobs. Especially in some fundamental, safety-net fields, ensuring a "human-in-the-loop" can be seen as a crucial guarantee for safeguarding the right to human employment.Real-World References for the Quota System
The quota system is not a baseless fantasy but an employment solution and resource allocation method already operating in our society, though not yet fully recognized as such.Take our country's "Employment Quota System for the Disabled" as an example. According to the "Regulations on the Employment of Persons with Disabilities," the proportion of disabled employees arranged by an employer must not be lower than 1.5% of the total number of in-service employees. If a unit fails to meet the specified proportion, it needs to pay a Disabled Employment Security Fund.The payment standard is the average annual salary of in-service employees from the previous year, multiplied by a payment coefficient. If the actual proportion of disabled employees arranged is below 1%, the payment coefficient is 90%, roughly equivalent to the average social wage. If the arranged proportion is ≥1% but <1.5%, the payment coefficient is 50%.For those who exceed the quota ratio in arranging employment, corresponding reward mechanisms are also provided. For example, Hunan Province stipulates that for the portion exceeding the quota, a reward of six times the local monthly minimum wage standard will be given. Or units where the resettlement ratio reaches over 25% and the number of people meets the standard can enjoy tax incentives like immediate refund upon payment of value-added tax.From the measures and supporting plans of this policy, it's clear it's not just about solving the income problem for the disabled. The core is to encourage enterprises to absorb genuine employment and motivate them to implement real employment through a well-established reward and penalty mechanism.At the same time, the policy execution has sufficient flexibility to avoid enterprises falling into a rigid constraint or formalism. Perhaps some enterprises, due to a special production environment—like mining enterprises—might truly be unsuitable for arranging employment for the disabled. They can meet the quota requirement by paying the security fund.Through a rigid overall goal and flexible execution strategies, we can ensure the total supply of job positions for the disabled while allowing the enterprises undertaking the quota to decide how to complete the task based on their actual situation. This set of measures and systems can arguably serve as the best reference template for implementing a "Human Quota System" in the context of AI's employment impact.Furthermore, the quota system, as a resource allocation method, is not only practically applied in ensuring employment for the disabled. In the commercial realm, quota systems are not uncommon, such as the purchase-with-purchase or allocation system in the luxury goods sector.If you want to buy a rare, newly launched product, sorry, you must purchase other goods first—spend a certain amount on regular items to qualify for the limited edition. The limited edition is a scarce resource; you must meet the quota to obtain it. In the context of a human quota system, humans themselves are the "limited edition."When adopting AI technology on a large scale, sufficient human jobs must be secured before the system can officially go live. Regardless of the field an AI company or application enters, if you want to enter a market, you must first fulfill the "human quota."Some might wonder if, under the onslaught of AI technology, a future development into a human quota system would signify entering a tragic world where "the value of human labor is no longer determined by market demand but by institutional pity."But looking at it from another angle, the "Human Quota System" is not about pity. Rather, it is about establishing, through legal form, the right for people to earn income by participating in labor. The core of the "Human Quota System" is not to demean people but to use legal and institutional constraints to guarantee that "human existence itself is an irreplaceable, scarce value."To address the employment impact of AI, there are multiple potential measures and proposals, including taxing AI and promoting Universal Basic Income (UBI). The latter, in particular, has gained significant recognition and support from many Silicon Valley tech entrepreneurs.For instance, in its 2026 white paper, "Industrial Policy in the Intelligent Age: A Human-Centred Vision," OpenAI proposed taxing automated labor to fill the funding gap in social security systems, and simultaneously establishing a "Public Wealth Fund" capitalized by governments and AI companies, allowing every citizen to share in the economic growth brought by AI.Whether through taxation, funding, or UBI, although the advocated forms differ, the core follows the same logic: if AI takes away your job, then we tax the AI to subsidize you. The core of this solution is the direct redistribution of wealth, rather than safeguarding employment.Like the "Human Quota System," Universal Basic Income seems like a completely new proposition, but some references can be found. For example, certain countries with abundant natural resources or those that entered a high-welfare state early have social security measures that are close to or even meet UBI standards.However, these countries share some common characteristics: they have small populations and can rely on a single resource or industry to ensure a high standard of living for all citizens. For more populous countries, especially developing economies, or regions where resources and welfare conditions are not yet in place, pure taxation or UBI faces multiple difficulties and challenges.Firstly, the subject of taxation: who should be taxed? How are boundaries defined? Is it chip companies, model companies, or also application providers? If AI becomes ubiquitous in every industry, does it necessitate universal taxation? Would universal taxation accelerate industrial relocation?Secondly, the depth of funds: how much can be collected? If unemployment due to AI proliferation potentially reaches hundreds of millions, then even if high tax rates are levied on a few industries and enterprises, the amount each person receives might be a drop in the bucket.The deeper issue with implementing UBI lies in how this plan reshapes the connection between people and society—it transforms people from workers into "beneficiaries." In the traditional social contract, everyone exchanges labor (physical or mental) for pay, thereby gaining social status and dignity. Work is not just a means of livelihood but an anchor for participating in the social division of labor and proving self-worth.Once UBI becomes mainstream, this connection is severed. In "Homo Deus," Yuval Noah Harari proposed a concept worth watching out for: the "useless class." This is not a moral judgment about laziness or diligence but a cold judgment from economics and political science. These people might rely on UBI to maintain basic survival, but they no longer play any role in the economy and production. UBI might eliminate the panic of survival, but it comes at the cost of human independence and dignity.Therefore, simply handing out money cannot solve all the problems brought by AI. It's just a painkiller, not a cure. The real challenge is that, beyond UBI, we need to redefine value and meaning. We need to build a new social ecosystem where caring for family, volunteering, lifelong learning, and even pure entertainment and socializing can be seen as valuable contributions, not merely labeled as being idle or not engaged in proper work.The "Human Quota System," precisely by using legal means to channel ordinary people's economic and social participation into the rule of law. Just as individuals with disabilities who gain job opportunities through the employment quota system don't need to be personally grateful to the company; your benefit doesn't stem from the charity of a business owner but from legal protection, and law is the concentrated embodiment of societal consensus. Just like the legal rights to weekend rest, paid annual leave, and minimum wage guarantees.You don't feel guilty towards your boss for not working on the weekend because it's a legal right. Similarly, under the human quota system, once an industry seeks to implement large-scale AI replacement, it must guarantee a sufficient "human quota," meaning hiring a certain percentage of human practitioners. This means that humans are not the objects of charity but the prerequisite for entry.Under the UBI model, humans are "kept alive"; under the human quota model, humans are "needed." These different positions determine the difference in human subjectivity in the age of AI.The Real Test of "People-Centricity"
In the development of technology, "putting people first" is a concept often mentioned. But truly achieving "people-centricity" encompasses not only the technical dimension but also the strategic dimension.Technically, people-first design means a more user-friendly interface, more convenient operation, and a smoother interactive experience. People-centricity at the technical level is very easy to judge; the users are the best arbiters. However, people-centricity at the strategic level is much harder to visualize intuitively. It often involves more complex factors, even contradictory considerations.Just like in the film and television industry, although AI replaced jobs like extras, costume designers, and makeup artists, it also provided new opportunities for prompt engineers and card drawers. Using AI to replace stars and extras certainly can improve efficiency and reduce costs. Whether users like it or not can be left to the market to decide. But behind the large-scale application of AI, there is inevitably an erosion of the subjectivity of "people" and an all-around shock to the content ecosystem.Of course, under the historical trend of AI deeply integrating into the content industry, completely rejecting AI is also inadvisable. Therefore, people-centricity at the strategic level rarely has a single, clear path; it requires every entrepreneur, every product manager, and everyone making decisions with AI to choose and answer for themselves.But a general principle is: Is AI technology development used to replace people or to enhance them? Does technological progress make people more at ease or more anxious? The proposal of the human quota system is also an attempt and exploration of "people-centricity" in addressing employment issues.Under the impact of AI, whether to believe in the market's natural selection, adopt the UBI of direct welfare safety nets, or explore a human quota system, this is not just an economic and social policy choice, but a profound interrogation of the value proposition of "humanity" in the intelligent age.At this crossroads of technological change, technology itself has no good or evil, but the people who use it do. If we choose to use AI to squeeze the living space of humans, then anxiety will be the inevitable endpoint. But if we choose to use AI to extend human wisdom and ability, to put people at ease and amplify people, then technology will become a ladder promoting the leap of human civilization.Therefore, from ensuring employment to sustainable development, from personal dignity to exploring a new social contract, as an explorative direction to address AI-driven job displacement, the human quota system is a response plan worth discussing. Perhaps it is not perfect, and maybe the large-scale employment displacement we fear will not happen, but beyond assumptions, we also need to proactively formulate a deterministic buffer mechanism.* This article is an entry for the Tencent Research Institute's "Spring of Long Writings" activity.