社会与工程心理学研究室知识库

人工智能正在逐渐挑战人类社会的传统构成、互动方式甚至社会秩序，在经济、安全等方面带来的威胁不容忽视。如何适应智能社会，如何管理人工智能的发展，是步入智能社会的人类所面临的极其重要的课题。探索人工智能威胁与国家认同之间的关系，有助于揭示人们如何通过群体认同这种具有人类进化特质的心理策略应对人工智能所带来的风险，从科技形成社会生态环境的视角，补充群体认同理论框架，并为人工智能发展的管理以及智能社会背景下维持国家社会稳定相关的政策、法律法规的制定，提供心理学依据。

人工智能威胁，指在人类、社会和个人层面上人工智能对工作、安全、资源和生存等利益造成的威胁。国家政府被人们认为是人工智能技术发展和风险管理的行为主体，对人工智能威胁的感知增强，很可能导致政府信任水平的下降，进而带来国家认同水平降低。基于此，本研究将基于三个研究，分别通过公开数据集、在线问卷调查和线上行为实验，探讨人工智能威胁对国家认同的影响及政府信任在其中的心理机制。三个研究间具有补充和递进的关系，不仅提供了在不同国家文化背景下的结果稳定性，除了调查研究对相关关系的探讨，还通过行为实验研究进一步检验了二者的因果路径。

研究 1 基于全球大样本调查的跨文化公开数据集（包含 20 个国家和地区），采用了其中反映人工智能威胁感知的二分变量指标，初步探讨了人工智能威胁和国家认同的关系及政府信任的中介作用。多层线性模型分析结果表明，在国际范围内，相比对人工智能威胁感知较弱的个体，对人工智能威胁感知较强的个体会表现出更低的政府信任水平，进而表现出更低的国家认同水平，该结果具有跨文化稳定性。

研究 2 在对人工智能威胁、政府信任和国家认同变量具体化测量的基础上， 通过线上问卷社会调查（有效样本 541 人），探讨了人工智能威胁与国家认同之间的关系，及政府信任的中介作用。研究结果复证了研究 1 的发现，人工智能威胁与国家认同存在显著的负向关系，政府信任在其中起到中介作用，人工智能威胁越强，政府信任水平越低，国家认同水平越低。

研究 3 采用线上行为实验（有效被试 600 人）探究人工智能威胁与国家认同的因果关系。实验通过操纵虚拟的国家社会情境中人工智能与个体间的关系，启动了不同强弱程度的人工智能威胁，探讨了人工智能威胁是否通过政府信任影响国家认同。研究发现人工智能威胁对国家认同存在显著的负向影响，对于强人工智能威胁组的被试，其国家认同水平显著低于弱人工智能威胁组。同时，研究还发现了政府信任显著的中介效应，人工智能威胁的增强通过降低政府信任水平，引起国家认同水平的下降。该研究为理论路径的因果逻辑提供了重要证据。

以上研究，从相关到因果，揭示了人工智能威胁对国家认同的影响及其心理机制。本研究在理论上丰富了群体认同理论框架，发现了人工智能威胁与国家群体情感的联系，并提出了政府信任在其中的心理路径；在实践上也为理解人工智能发展带来的社会影响、完善智能社会背景下维持国家社会稳定相关的政策、法律法规的制定提供心理学证据的支持。

Artificial intelligence (AI) is gradually challenging the traditional structures, interactions, and even social orders of human society, posing threats in economic and security realms. Adapting to the intelligent society and manage the development of AI are crucial challenges confronting humanity in the era of intelligence. Exploring the relationship between AI threats and national identification helps reveal how people employ the psychological strategy of group identity, a trait of human evolution, to cope with the risks posed by AI. This study, from the perspective of technology shaping the socioecological environment, supplements the framework of group identity theory and provides psychological evidence for the management of AI development and the formulation of policies, laws, and regulations relevant to maintaining national social stability in the context of an intelligent society.

AI threats refer to the threats posed by AI to interests such as work, security, resources, and survival at the levels of individuals, society, and humanity. An enhanced perception of AI threats may lead to a decline in the perception of national collective values. Consequently, this can result in decreased levels of government trust and national identification. Based on this, the present study conducts three investigations using publicly available datasets, online surveys, and online behavioral experiments to explore the relationship between AI threats and national identification, as well as the psychological mechanisms of government trust therein. These investigations complement and build upon each other, providing not only stability of results across different cultural backgrounds but also examining the causal pathways between AI threats and national identity through behavioral experiments.

Study 1, based on a cross-cultural publicly available dataset derived from a global large-scale survey (including 20 countries and regions), employs a binary variable indicator reflecting the perception of AI threats to preliminarily explore the relationship between AI threats and national identification, as well as the mediating role of government trust. Multilevel linear model analysis reveals that, internationally, individuals perceiving stronger AI threats exhibit lower levels of government trust, subsequently leading to lower levels of national identification, a result demonstrating cross-cultural stability.

Study 2, based on structured measurements of AI threats, focuses on exploring the relationship between AI threats and national identification, as well as the mediating role of government trust, through a questionnaire survey (with an effective sample of 541 respondents). This study enriches the content of AI threat variables from the perspectives of the human world, national society, and individual life. The results replicate the findings of Study 1, showing a significant negative relationship between AI threats and national identification. Additionally, there is a significant mediating role of government trust, where higher AI threats correspond to lower levels of both government trust and national identification.

Study 3 employs online behavioral experiments (with 600 valid participants) to investigate the causal relationship between AI threats and national identification. By manipulating the relationship between AI and individuals in virtual national social contexts, experiments explore how AI threats affect national identification via government trust. The study reveals a significant negative impact of AI threats on national identification, with markedly lower levels of national identification observed in the strong AI threat group compared to the weak threat group. Specifically, the enhancement of AI threats leads to a decrease in government trust, causing a decline in national identification levels. This study provides important evidence for the causal logic of theoretical pathways.

The aforementioned studies, from correlation to causation, reveal the impact of AI threats on national identification and its psychological mechanisms. These findings not only enrich the framework of group identity theory theoretically but also uncover the connection between AI threats and national collective emotions, as well as the psychological pathway of government trust. Moreover, they provide psychological evidence to understand the social impacts of AI development and to improve the formulation of policies, laws, and regulations relevant to maintaining national social stability in the context of an intelligent society.