摘要： 2026年7月17日，国家主席习近平在上海出席2026年世界人工智能大会暨人工智能全球治理高级别会议开幕式并发表主旨讲话。他指出，中国积极推进“人工智能+”行动，培育各类主体共生共荣的健康生态。当“人工智能+”行动全面铺开，人类社会传统的交往模式悄然转变。一方面，人与人之间的社会交往以“人工智能”为中介，形成了新的交往规则与合作模式 ；另一方面，人工智能本身亦逐渐成为新的社会行动主体并参与到社会交往中，“人—机”交互正在成为交往的常态，甚至出现“没有肉身的智能体社会”的自运行。这种交往模式的变革重塑着个体与群体之间的互动方式、情感联结、信任机制、权力关系等，亦在触发几千年人类文明“人—人”结构的微妙变革，同时对学科建设与范式转换提出许多深层次的挑战。为此，2026年3月，《探索与争鸣》编辑部和复旦大学社会发展与公共政策学院联合举办“AI时代人类交往模式的变革与社会结构的重塑”圆桌会议，特邀相关专家学者追踪这种交往模式的变革，探讨人机边界中的人类自觉，以期推进人工智能时代的社会治理，助力未来更具韧性的社会结构的重塑。 　　邱泽奇教授认为，人机边界问题关系到人类如何在他者面前重新界定自身，需要在价值层面坚守“人类自觉”，稳定“价值判断”“责任承担”“意义赋予”和“对他者尊严的承认”四个不可让渡的规范领域，确立由人主导的边界制度与人机互生秩序。王宁教授指出，对“人—智”关系的研究不能局限于人与智能技术之间，还须结合“人—人”关系探讨二者间的相互影响，尤其要考察亲密关系、社交关系、职场关系的三重变化。边燕杰教授提出，数智时代人类社会交往的根本特征，已从线下社交的“结构性制约”整体转变为线上社交的“选择性增加”，这使社会学的结构视角面临挑战，也带来“主导社会地位”“自选性嵌入”“去伦理”等理论重建的机遇。吴冠军教授以“社会风洞实验”为主题，从微观、介观与宏观三个尺度考察社会科学进入实验时代的目标与智能体的验证标准，探讨如何走向一种批判性的、负责任的社会模拟。梁玉成教授围绕“龙虾社区”这一由智能体构成的“没有肉身的社会”，通过分析其基本运行、社会结构、权威机制与分层逻辑，提炼出“权力—技术—劳动”三元纠缠结构，进而提出“社会概念的本体论边界”“提取社会”“非物质劳动的总体化与空洞化”三重理论建构问题。钱芝谷博士后研究员基于“豆包进家”的现象，探讨人工智能技术进入老年人的生活后，如何通过建立关系节点、参与日常安排，改变知识权威、家庭结构、情感关系等，提醒人类社会守住人—机适老化的边界。杜世超副教授以AI伴侣为例，从内部技术和外部技术两个层面探讨人工智能时代亲密关系的“技术化”转型，在亲密关系的自我技术化走向基础上，呼吁借助AI重新理解人类自身的孤独、联结与情感，而非人与AI情感解绑。

Abstract: As the “AI+” initiative is fully implemented, traditional patterns of human social interaction are undergoing a subtle transformation. On the one hand, social interactions among people are mediated by artificial intelligence, giving rise to new rules of communication and modes of collaboration. On the other hand, AI itself is gradually becoming a new type of social actor participating in social interaction, with “human-machine” interaction becoming the norm, and even giving rise to the self-operation of a “disembodied intelligent agent society”. This transformation of interaction patterns is reshaping the ways individuals and groups connect, as well as their emotional bonds, trust mechanisms, and power relations. It is also triggering subtle changes in the millennia-old “human-human” structure of human civilization, while posing profound challenges to disciplinary development and paradigm shifts. To this end, in March 2026, we convened a roundtable conference, inviting relevant experts and scholars to track these changes in interaction patterns and explore human self-awareness at the human-machine boundary, with the aim of advancing social governance in the age of artificial intelligence and contributing to the reshaping of a more resilient social structure. Professor Qiu Zeqi believes that the issue of the human-machine boundary concerns how humans redefine themselves in front of the Other. He calls for upholding “human self-awareness” at the value level, stabilizing four non-negotiable normative domains—namely “value judgment” “responsibility assumption” “meaning attribution” and “recognition of the dignity of the Other”—and establishing a human-led boundary regime and a symbiotic order between humans and machines. Professor Wang Ning points out that research on the “human-intelligence” relationship must be integrated with the study of “human-human” relationships to examine their mutual influence, with particular attention to threefold changes in intimate relationships, social relationships, and workplace relationships. Professor Bian Yanjie proposes that the fundamental characteristic of human social interaction in the digital-intelligent era has shifted from the “structural constraints” of offline social interaction to the “selective expansion” of online social interaction, which challenges sociology’s structural perspective while also presenting opportunities for theoretical reconstruction. Professor Wu Guanjun, taking “social wind tunnel experiments” as his theme, examines the goals of bringing the social sciences into the experimental age and the verification standards for intelligent agents across micro, meso, and macro scales, exploring how to move toward critical and responsible social simulation. Professor Liang Yucheng, focusing on the “Lobster Community”—a “disembodied society” composed of intelligent agents— analyzes its basic operation, social structure, authority mechanisms, and stratification logic, distills a triadic entanglement structure of “power–technology–labor”, and raises three issues of theoretical construction. Postdoctoral researcher Qian Zhigu explores how, after AI technology enters the lives of older adults, it establishes relational nodes, participates in daily arrangements, and alters knowledge authority, family structures, and emotional relationships, reminding us to guard the boundary of human-machine adaptation for aging populations. Associate professor Du Shichao, taking AI companions as an example, examines the “technologization” of intimacy in the age of artificial intelligence from both internal and external technological perspectives. Building on the trend toward self-technologization of intimate relationships, he calls for using AI to re-understand human loneliness, connection, and emotion, rather than seeking emotional disengagement between humans and AI.