In the era of artificial intelligence, we need to establish and improve a universal social security system to ensure that productivity is shared and workers are protected. This is the essential aspect of addressing the impact of artificial intelligence on employment. We can argue based on several characteristics exhibited by artificial intelligence. Firstly, the creative destruction of artificial intelligence is a whole, and without 'destruction', there can be no 'creation'. Like any advanced technology, artificial intelligence is endowed with the ultimate expectation of enhancing people's well-being, but technology and well-being are ultimately not the same thing, and there are differences in motivation and functional boundaries between the two. For example, the function of artificial intelligence in improving productivity and promoting economic growth ultimately lies in its creative and destructive nature, which are essentially two sides of the same coin. Secondly, there is no such thing as a "trickle down effect" in the world, but rather a "synthesis fallacy" often arises. For major national governments, the development of artificial intelligence has become a key lever for inter country competition to occupy the high ground and form economic growth points, resulting in significant subsidies and other policy incentives. For investors, large-scale investment in this field is not for charity or public welfare, but rather for expected high return venture capital. For technology innovation enterprises, whether they can seize the initiative in this field is not only related to their competitiveness, but also seems to be a watershed of life and death. For business entities that apply technology, their starting point is naturally to reduce staff and increase efficiency, reduce costs and increase profits. The motives and pursuits of all the parties mentioned above not only have great power, but also manifest as: no matter what reason they choose to observe, they will not receive a response from partner companies and competitors, but will only leave themselves behind. Since market entities cannot proactively control the potential risks and impact effects of artificial intelligence, government regulation and control measures cannot be absent or delayed. Finally, the difficulty of identifying beneficiaries of social security is becoming increasingly prominent. The social security system strives to protect the truly disadvantaged population while avoiding the practice of "raising lazy people" as much as possible. However, in the era of artificial intelligence, the operating conditions of this model will undergo fundamental changes. Under the automation technology empowered by artificial intelligence, the relationship between workers, job positions, and social security is undergoing and will continue to undergo changes that subvert traditional concepts. This change requires corresponding shifts in cognitive paradigms and policy guidance. An increasingly evident phenomenon is that society as a whole is becoming increasingly difficult to predict in advance: which skills of workers can complement artificial intelligence and be expected to be enhanced by new technologies; What kind of worker skills exactly compete with artificial intelligence, making workers who master these skills the target of being replaced by artificial intelligence in their positions. It can be seen that the mismatch of human resources in the labor market is not the fault of any individual worker. And once workers understand this truth from their experience of encountering employment shocks, their rational choice is to continuously receive training and re education, sometimes even requiring some time for observation and waiting. For society, a better way is to provide a safety net in a more inclusive manner, allowing affected workers to try and wait. In general, if the individual employment intention, reasons for unemployment, and labor participation status of workers cannot be identified with certainty, the social security system should be more inclusive in order to leave more room for labor market behavior. The changes in China's labor market and income distribution pattern have led to new institutional demands. After the transformation of the main contradiction in the labor market, how to respond more targetedly to structural employment contradictions; How to promote equal sharing of higher productivity brought by technological progress among all workers; How to effectively protect the rights and interests of workers and the well-being of the people in a risky technological change environment. To address these issues, it is necessary to deepen our understanding of the development laws of artificial intelligence under the guidance of new development concepts, and appropriately determine the starting point and foothold of institutional construction. It is urgent to carry out a paradigm shift based conceptual update. Although the two systems of social security and social welfare have fundamental similarities, namely achieving social protection goals through social solidarity and sharing development achievements through social coordination, we can also see some obvious differences between the two. Firstly, there are certain differences in the institutional concepts and principles relied upon. Social security only provides guarantees for the most basic needs of defined populations, while social welfare determines a universal standard for basic human rights based on the level of social necessity, and makes social welfare arrangements without discrimination. Therefore, the scope of public services provided by social welfare should be significantly larger than that of social security, aimed at meeting the needs of various stages of the life cycle. At the same time, the level of social welfare protection is also higher, and with the continuous improvement of productivity, it is rising rapidly. Secondly, there are fundamental differences in the way public services are provided. Many social security programs typically require identifying targets through livelihood surveys or similar methods to ensure the protection of "eligible" populations. The social welfare system provides public services in a universal manner, targeting all citizens or groups with a common characteristic as a whole. Finally, fiscal philosophy and budget discipline are different. The social security system tends to adhere to traditional fiscal concepts, pursuing increased revenue and reduced expenses, fixed expenditures based on revenue, and current balance. The social welfare system considers fiscal "balance" from a longer-term perspective, emphasizing both the significance of expenditures for social protection and the role of expenditures in cultivating human capital, with a focus on better economic performance and rising fiscal revenue and expenditure. The enormous productivity improvement expected from artificial intelligence can ultimately break the absolute constraints of resources and provide a solid material foundation for building a higher-level welfare state. On this basis, China's social security system will accept more social welfare concepts and practices, focusing on enhancing inclusiveness and covering the entire life cycle, to provide more extensive, sufficient, and equal basic public services for all residents. Whether it is adopting a low-key expression and a gradual transition, substantially shifting towards a more inclusive basic public service system, or making a high-profile declaration of the goal of building a welfare state with Chinese characteristics, we cannot avoid answering questions about whether the national financial resources can support it and how to raise funds. This does not mean that the principle of "doing one's best and acting within one's capacity" can be downplayed. However, at a higher level of economic development and with the expectation of significantly improving productivity through artificial intelligence, the government's "best effort" in its work should increase day by day, and the "capacity" of fiscal capacity has also reached a time of rising tide. Larger scale investments are not always manifested as an increase in the total amount of funds, but rather as adjustments to the expenditure structure in more cases. The approach involved here is to shift from "investing in things" to "investing in people". From the perspective of people's livelihood needs and international comparison, there is still room for adjustment in the financial resources and expenditure structure of the Chinese government. Research also shows that material capital investment in the Chinese economy is facing continuous diminishing returns, while investment in people will maintain a high rate of return in the long run. More and higher quality social welfare supply is essentially equivalent to the state transforming material capital facing diminishing returns into human capital with increasing returns in the form of institutions. For a long time, public policy concepts have adopted a cautious and careful attitude to prevent the "growth while distribution" model from exceeding its limits and evolving into a "distribution first, growth assistance" model, falling into a situation of living beyond the means and even falling into the trap of populist policies. It can be said that this cautious attitude, especially the rejection of the "allocate first, help growth" model, has long had sufficient reasons. However, due to at least two important changes and the resulting new situations, there is an increasing need for a renewed understanding of such issues. In other words, we urgently need to carry out a paradigm shift based conceptual update. The first change is that the main constraining factor for China's economic growth has shifted from the supply side to the demand side, with household consumption increasingly becoming the decisive driving force for growth, and the latter's boost relies on more inclusive social security. The second change is that the development of artificial intelligence brings us a great possibility of significantly improving productivity, and the necessity of pre allocating this larger production rate cake can be supported by economic theory. In this sense, various existing institutional models and policy measures, such as the "unconditional basic income" distributed to all citizens, the "universal basic pension" that does not require payment and qualification recognition, and the "living wage" determined based on the guarantee of necessary social quality of life, may not be suitable for wholesale adoption, but they all have beneficial elements for reference. (Xinhua News Agency) Author: Cai Fang (Member of the Faculty of Chinese Academy of Social Sciences)