Innovation in Scene Methods for Ideological and Political Education in the Era of Large Models

The scene of ideological and political education is essentially a deep integration of “field” and “setting” mediated by educational technology, a mode of education in which the “field” of connective relations carries the “setting” of value transmission. The arrival of the era of large models has brought about a scene revolution in ideological and political education. It has not only transformed the connective “field” of ideological and political education, endowing it with characteristics such as the interconnection of all things, cross-border integration, and shared co-creation, but has also made the educational “setting” increasingly personalized, integrated, and emergent. The breakthrough development of large-model technology drives the intelligent transformation of the scene of ideological and political education, opening new pathways and expanding new prospects for the innovative development of ideological and political education. The intelligent transformation of the scene of ideological and political education is inseparable from the innovation and application of scene methods. In the era of large models, following the logic of inquiry of “scene narrative—scene analysis—scene evolution,” a systematic examination of the scene methods of ideological and political education holds positive practical significance for improving the efficiency, effectiveness, and impact of scene-based education in ideological and political work.

思想政治教育场景本质上是教育技术中介下“场”与“景”的深度融合，是以连接关系之“场”承载价值传播之“景”的育人方式。大模型时代的到来带来了思想政治教育的场景革命，它不仅变革了思想政治教育的连接之“场”，使其呈现出万物互联、跨界融合、共享共创等特征，而且使育人之“景”变得日益个性化、整合化与涌现化。大模型技术的突破性发展推动思想政治教育场景的智能化变革，为思想政治教育的创新发展开辟了新路径，开拓了新前景。思想政治教育场景的智能化变革离不开场景方法的创新及应用。大模型时代遵循“场景叙事—场景分析—场景进化”的追问逻辑，对思想政治教育的场景方法作出系统的考察，对提升思想政治教育场景育人的效率、效益与效果有着积极的现实意义。

I. Innovation in Scene Narrative Methods for Ideological and Political Education in the Era of Large Models

一、大模型时代思想政治教育场景叙事方法创新

The arrival of the era of large models has driven a transformation in the scene narrative methods of ideological and political education, concentratedly manifested in the dimensions of subject, time, space, and ideology.

大模型时代的到来推动了思想政治教育场景叙事的方法变革，并集中表现在主体、时间、空间与意识形态等维度。

1. Subject Narrative: From Informational Narrative to Nodal Narrative

1.主体叙事：从信息化叙事到节点化叙事

The arrival of the era of large models not only presents the scene of ideological and political education in the form of an information system but also causes the subjects of the educational scene to exist in informational and nodal modes of information, bringing about a deep transformation in subject narrative. First, informational narrative. The subjects of the educational scene empowered by large models exist both as receptive informational relations and as disseminative informational practices, providing an ontological foundation for the informational transformation of subject narrative. As receptive informational relations, the subjects of the educational scene are “information bodies that receive, process, and understand existing information, and create and transmit innovative information.” Driven by functions such as information adaptation and information synthesis in large models, this not only promotes innovation in narrative methods such as information archive narrative, digital image narrative, and digital imagination narrative, but also, through their interdependence, mutual influence, and co-action, manifests the anthropological characteristics of scene narrative. As disseminative informational practices, large models not only bring customized and functional practical changes to the information dissemination of scene subjects, thereby enhancing the precision and experiential quality of scene subject narrative, but also, in the systematic integration of information use, sharing, and creation, enhance the comprehensibility, interpretability, and emergence of scene narrative, promoting the organic construction of a scene narrative ecology. Second, nodal narrative. The subjects of the educational scene empowered by large models exist in the educational scene as an information system in forms such as communication network nodes, social network nodes, and service network nodes. As communication network nodes, scene subject narrative manifests in forms such as critical narrative, guiding narrative, and integrative narrative. Among these, critical narrative is embodied in the scene subjects’ examination and reflection on scene information based on large models; guiding narrative is embodied in the information dissemination, persuasion, and guidance carried out by scene subjects acting as official media and opinion leaders, relying on large models; integrative narrative is embodied in the self-organizing narrative of the educational scene system activated by large models and the information integration narrative of scene information producers. As social network nodes, large models can establish connective relationships among scene subjects through the timely appearance of information, establish interactive relationships through scene subject performance and the construction of learning communities, and consolidate subjects’ ideological consensus and activate collective wisdom through the construction of scene identification, trust, and dialogue relationships. As service network nodes, large models, through real-time tracking of user status, analysis of living habits, and creation of social atmosphere, truly realize a service narrative of real-time connection, personalized adaptation, and contextual integration.

大模型时代的到来不仅使思想政治教育场景以信息系统的形态呈现出来，而且使教育场景主体以信息化与节点化的信息方式存在，它带来了主体叙事的深层变革。一是信息化叙事。大模型赋能下的教育场景主体既作为感受性信息关系存在，又作为传播性信息实践存在，它们为主体叙事的信息化变革提供了存在论基础。作为感受性信息关系，教育场景主体即“接收、加工、理解已有信息，创构和发送创新性信息的信息体”。它在大模型的信息适配、信息综合等功能的驱动下，不仅推动了信息档案叙事、数字图像叙事、数字想象叙事等叙事方法的创新，而且以其相互依存、相互影响、共同行动的交互性彰显了场景叙事的人类学特性。作为传播性信息实践存在，大模型不仅为场景主体的信息传播带来了定制化与功能化的实践变革，进而增强了场景主体叙事的精准性与体验性，而且在信息使用、共享、创构的系统整合中增强了场景叙事的理解性、诠释性与涌现性，推动了场景叙事生态的有机建构。二是节点化叙事。大模型赋能下的教育场景主体在作为信息系统的教育场景中以传播网络节点、社会网络节点以及服务网络节点等形态存在。作为传播网络节点，场景主体叙事表现为批判性叙事、引导性叙事、整合性叙事等形式。其中，批判性叙事体现为场景主体基于大模型的场景信息考辨与反思；引导性叙事体现为作为官方媒介与意见领袖的场景主体依托大模型展开的信息传播、说服与引导；整合性叙事体现为大模型激活下教育场景系统的自组织叙事与场景信息生产者的信息整合叙事。作为社会网络节点，大模型能够通过信息的及时出场建立场景主体连接关系，通过场景主体展演与学习共同体的建构建立场景交互关系，通过场景认同、信任与对话关系的建构凝聚主体思想共识、激活群体智慧等。作为服务网络节点，大模型通过对用户的实时状态追踪、生活习惯分析以及社交氛围营造，真正实现了实时对接、个性适配、情景交融的服务叙事。

2. Time Narrative: From Real-Time Narrative to Customized Narrative

2.时间叙事：从实时叙事到自定义叙事

Large models activate the virtual time attribute of the scene of ideological and political education, placing scene narrative within the dual construction of real-time narrative and customized narrative. First, real-time narrative. Large models reconstruct the virtual culture in which the scene of ideological and political education is situated. By activating the virtual attributes of the educational scene, they place scene narrative in a state of “always immediate” and “waiting for you to activate.” “The culture of real virtuality, accompanied by the electronically integrated multimedia system, contributes in two different ways to the transformation of time in our society, namely simultaneity and timelessness.” [1]1 Manuel Castells, *The Rise of the Network Society*, trans. Xia Zhujiu et al. (Beijing: Social Sciences Academic Press, 2006), p. 464. [The original English text is from Manuel Castells, *The Rise of the Network Society*, 2nd ed. (Oxford: Blackwell, 2000), p. 464.] Specifically, in the scene activities of ideological and political education empowered by large models, virtual time, as a “transformation of time,” not only achieves the instantaneous presentation of social events and virtual culture through the instantaneous flow of information, but also, as “eternal-instantaneous time,” makes time without time a reality. Virtual time is said to be eternal “because it circulates repeatedly through the entire sequence of cultural representations”; it is said to be instantaneous “because each arrangement, each specific ordering, depends on the context and intention triggered by any given cultural construction.” [2]2 Ibid., p. 465. Second, customized narrative. Large models activate the customizability of virtual time through the reconstruction of time experience, providing a new practical basis for the time narrative of the educational scene. Furthermore, large models place educational scene activities within customized time experiences of entertainment, interactivity, and value, thereby “seizing” users’ other time or increasing users’ scene immersion time. Among these, the customized time experience of entertainment manifests as solving the lag problem in scene subjects’ information queries, solving various “pain point” problems such as the difficulty of real-time recording of inspiration information, and the real-time supply, enrichment, and satisfaction of information for spiritual space and free time. The customized time experience of interactivity manifests as psychological or ideological resonance in product scene interaction, activity scene interaction, and community instant interaction, placing educational subjects’ scene participation in a state of total immersion. The customized time experience of value manifests as the activation of deep-seated needs of educational scene subjects through “social currency” such as the satisfaction of needs for talking points, expression, image, and values: “If sharing a certain content can improve your value in the eyes of others, then this content, like currency, buys their newfound respect for you from the outside world.” [3]3 Jonah Berger, *Contagious: Why Things Catch On*, trans. Liao Jianqiao (Beijing: Publishing House of Electronics Industry, 2014), p. 37. [The original English text is from Jonah Berger, *Contagious: Why Things Catch On* (New York: Simon & Schuster, 2013), p. 37.]

大模型激活了思想政治教育场景的虚拟时间属性，使场景叙事处于实时叙事与自定义叙事的双重建构之中。一是实时叙事。大模型重构了思想政治教育场景置身其中的虚拟文化，以其对教育场景的虚拟属性激活，使场景叙事处于“永远即时”与“等你激活”的状态，“真实虚拟的文化伴随了电子整合的多媒体系统，以两种不同的方式促成我们社会里时间的转化，亦即同时性（simultaneity）与无时间性（timelessness）”。具体来讲，在大模型赋能的思想政治教育场景活动中，虚拟时间作为“时间的转化”，不仅在信息的瞬间流转中实现了社会事件和虚拟文化的即时性呈现，而且作为“永恒—瞬间的时间”，使无时间之时间成为现实。其中，之所以说虚拟时间是永恒的，“因为它循环往复于整个文化表现的序列中”；之所以说虚拟时间是瞬间的，“因为每个安排，每个特定的排序根据任何既定的文化建构所引发的脉络与意图而定”。二是自定义叙事。大模型通过时间体验的重构激活了虚拟时间的自定义性，它为教育场景的时间叙事提供了新的实践基础。进而言之，大模型使教育场景活动处于娱乐性、交互性与价值性的自定义时间体验之中，并由此“抢占”用户的其他时间或增加用户的场景沉浸时间。其中，娱乐性的自定义时间体验表现为对教育场景主体信息查询的滞后性问题的解决，对灵感信息的实时记录困境等各类“痛点”问题的解决，精神空间与空闲时间的实时信息供给、丰富与满足，等等。交互性的自定义时间体验表现为产品场景交互、活动场景交互与社群即时交互中的心理或思想的同频共振，它使教育主体的场景参与处于全身心地沉浸状态之中。价值性的自定义时间体验表现为谈资、表达、形象、价值观需要满足等“社交货币”对教育场景主体深层次需要的激活，“如果分享某项内容可以提高你在别人眼中的价值，那么这条内容就像货币一样从外界买来对你的刮目相看”。

3. Space Narrative: From Mass Communication to Social Integration

3.空间叙事：从泛众传播到社会整合

Large models have reshaped the spatiality of the scene of ideological and political education, causing it to manifest a trend toward pan-massification and integration. This has brought about a spatial narrative revolution of pan-mass communication and social integration. First, pan-mass communication. Large models have created a ubiquitous communication network for scene narratives. They not only redefine the concept of the subject of the educational scene, transforming it from a "mass" or "segmented audience" form into a "pan-mass" form, but also shift the scene narrative model from mass communication and segmented communication toward pan-mass communication, that is, "personalized communication centered on each individual, provided to all people." [4]4 The authors are drawing on the concept of "pan-mass communication" (泛众传播) as developed in Chinese communication studies, which posits a shift from mass communication to a networked model where every individual can be both a sender and receiver of personalized content. Specifically, large models not only activate the centrality of the individual as a pan-mass, making the scene narrative a node-based narrative synthesis centered on the individual, but also enhance the self-immersive feeling of the "pan-mass," prompting them to consciously engage in the construction activities of personal space and social space. Furthermore, they promote the full, whole-process, and all-round development of the scene narrative. It should be noted that the pan-mass communication narrative not only has personalized attributes but also possesses shared and co-creative characteristics. On the one hand, the sharing transformation brought by large models has given rise to sharing services in the educational scene, providing an external driving force for pan-mass social interaction and co-creation based on collaborative cooperation. On the other hand, the precise information推送 and adaptation of large models activate the common needs of the "pan-mass," stimulating the internal motivation for cooperative creation among the "pan-mass." Second, social integration. The social integration of the educational scene empowered by large models is manifested in its function of dissolving and aggregating and its trend toward the unity of person and domain. On the one hand, large models highlight the social integration utility of the educational scene through their boundary-dissolving power and content-aggregating power. At the level of boundary-dissolving power, large models, by dissolving the boundaries between the virtual and the real and the boundaries of the body, promote the deep integration of an environmental narrative that "transforms things-in-themselves into things-for-us" and a situational narrative that "transforms things-for-themselves into things-that-transform-us," as well as the organic combination of a narrative of bodily absence and a narrative of attentional presence. At the level of content-aggregating power, the educational scene empowered by large models, through the integration of content such as the sphere of existence, values, cultural rituals, and game rules, not only achieves the effective elimination of dross content elements but also realizes the value identification and emotional resonance of the educational scene subjects with the scene content. On the other hand, the educational scene empowered by large models and the educational subjects gradually tend toward the unity of person and domain in scene immersion. In other words, large models, through their immersiveness, render the self as the scene subject formless, that is, prompting the educational subject to merge into the scene, becoming a part of the scene or becoming a scene medium for others.

大模型重塑了思想政治教育场景的空间性，并使其呈现泛众化与整合化的趋势，它带来了泛众传播与社会整合的空间叙事革命。一是泛众传播。大模型为教育场景叙事塑造了泛在化的交际网络，它不仅重新定义了教育场景主体的概念，使其从“大众”或“分众”形态转变为“泛众”形态，而且使场景叙事模式从大众传播、分众传播走向泛众传播，即“以每个个人为中心、面对所有人提供的个性化传播”。具体来讲，大模型不仅激活了作为泛众的个人的中心性，使场景叙事成为以个人为中心的节点化叙事综合，而且增强了“泛众”的自我沉浸感，促使其自觉地投入到个人空间与社会空间的建构活动中来，更推动了场景叙事的全员全程全方位开展。需要注意的是，泛众传播叙事不仅有着个性化的属性，而且拥有共享共创性。一方面，大模型的共享变革催生了教育场景的共享服务，它为泛众化的社交展开以及协同合作基础上的共同创造提供了外在驱动力。另一方面，大模型的信息精准推送与适配激活了“泛众”的共同需要，激发了“泛众”合作创造的内在动力。二是社会整合。大模型赋能下教育场景的社会整合在其发挥消融聚合功能与展现人域合一趋势中得以彰显。一方面，大模型以其边界消融力与内容聚合力凸显了教育场景的社会整合效用。在边界消融力层面，大模型通过对虚实边界与身体边界的消融，推进了“化自在之物为为我之物”的环境叙事与“化自为之物为化我之物”的情境叙事的深度融合，以及身体缺场叙事与注意力在场叙事的有机结合。在内容聚合力层面，大模型赋能下的教育场景通过对生存领域、价值观念、文化仪式、游戏规则等内容的整合，既实现了对糟粕内容元素的有效剔除，又实现了教育场景主体对场景内容的价值认同与情感共鸣。另一方面，大模型赋能下的教育场景与教育主体在场景沉浸中渐趋人域合一。换言之，大模型以其沉浸性使作为场景主体的自我趋于无形，即促使教育主体融于场景之中，成为场景的一部分或成为别人的场景媒介。

4. Ideological Narrative: From Narrative Content to Content Narrative

4.意识形态叙事：从叙事内容到内容叙事

Large models activate the ideological narrative function of the scene of ideological and political education in two dimensions: narrative content and content narrative. They not only satisfy the educational objects' need for mainstream ideology but also activate the educational objects' function of ideological recognition. First, the satisfaction of the need for mainstream ideology based on narrative content. Large models redefine the content of the narrative in the scene of ideological and political education, satisfying the educational objects' ideological needs through their comprehensiveness advantage and humanized characteristics. On the one hand, the educational scene empowered by large models manifests the comprehensiveness advantage of narrative content through characteristics such as knowledge integration, interdisciplinarity, and generativity. "It can partially imitate human thinking ability, but it has a comprehensiveness advantage over any individual, effectively promoting the cross-domain integration and cross-analysis of global knowledge, helping to discover potential connections between different fields, and potentially generating or stimulating new knowledge." As a carrier for bearing and disseminating mainstream ideological information, the more comprehensive educational narrative content provides a powerful and abundant information resource support for satisfying the educational objects' need for mainstream ideology. On the other hand, the educational scene empowered by large models has humanized characteristics. It can not only provide adaptive scene content according to the subjectivity and personalized needs of the educational objects to satisfy their need for mainstream ideology but also, based on the analysis and grasp of people's ideological conditions, generate scene content that fits the ideological needs of the educational objects and triggers their ideological resonance. Second, the activation of the function of ideological recognition based on content narrative. The content narrative of the educational scene empowered by large models has a dual strategy of "filtering" and "selecting," which strengthens the教化 function of mainstream ideology in the stimulation of the educational objects' ideological recognition. Specifically, the "filtering" strategy is manifested in the criticism and sublation of non-mainstream ideologies by large models on the basis of information identification; the "selecting" strategy is manifested in the enrichment and enhancement of mainstream ideological content by large models on the basis of information selection; the stimulation of ideological recognition is manifested in the educational scene relying on large models to "hail concrete individuals as concrete subjects." [5]5 This is an allusion to Louis Althusser's concept of interpellation (or hailing), in which ideology "recruits" individuals as subjects. The authors adapt this to describe how large models in the educational scene can call forth subjects who identify with mainstream ideology. Among them, "concrete individuals" refer to the specific educational objects situated in a complex and diverse ideological field; "concrete subjects" refer to the educational objects within the mainstream ideological field, that is, the educational objects who have formed identification with and consensus on mainstream ideology.

大模型在叙事内容与内容叙事两个维度激活了思想政治教育场景的意识形态叙事功能，既满足了教育对象对主流意识形态的需要，又激活了教育对象的意识形态承认功能。一是基于叙事内容的主流意识形态需要满足。大模型重新定义了思想政治教育场景叙事的内容，以其全面性优势与人性化特点满足了教育对象的意识形态需要。一方面，大模型赋能下的教育场景通过知识的整合性、交叉性、创生性等特性彰显了叙事内容的全面性优势，“它能部分模仿人类的思维能力，但又比任何一个人更具全面性的优势，能有效促进全球知识的跨领域整合、交叉分析，帮助发现不同领域间的潜在联系，有机会产生或激发出新的知识”。作为承载与传播主流意识形态信息的载体，更具全面性的教育叙事内容为满足教育对象的主流意识形态需要提供了雄厚有力的信息资源支撑。另一方面，大模型赋能下的教育场景有着人性化特点，它不仅能够根据教育对象的主体性与个性化需要提供适配性的场景内容以满足其主流意识形态需要，而且能够基于对人们思想状况的分析和把握，生成契合教育对象思想意识需要并引发其意识形态共鸣的场景内容。二是基于内容叙事的意识形态承认功能激活。大模型赋能下的教育场景内容叙事有着“过滤”与“选择”的双重策略，它在教育对象意识形态承认的激发中强化主流意识形态的教化作用。具体来讲，“过滤”策略表现为大模型在信息辨识基础上对非主流意识形态的批判与扬弃；“选择”策略表现为大模型在信息选择基础上的对主流意识形态内容的丰富与增强；意识形态承认的激发表现为教育场景依托大模型“把具体的个人唤为具体的主体”。其中，“具体的个人”指的是置身复杂多样的意识形态场中的具体教育对象；“具体的主体”即主流意识形态场中的教育对象，亦即对主流意识形态形成认同与共识的教育对象。

II. Innovation in Methods for Analyzing the Scene of Ideological and Political Education in the Era of Large Models

二、大模型时代思想政治教育场景分析方法创新

Large models have transformed the traditional methods for analyzing the scene of ideological and political education. Through data mining, network analysis, modeling and simulation, and computational experiments on the scene of ideological and political education, they have achieved the automatic extraction of scene semantics, the intelligent evaluation of scene value, the intelligent promotion of scene operation, and the virtual-real complementarity of scene improvement.

大模型变革了传统的思想政治教育场景分析方法，它通过对思想政治教育场景的数据挖掘、网络分析、建模仿真与计算实验，实现了场景语义的自动提取、场景价值的智能评估、场景运行的智能推进以及场景改进的虚实互补。

1. Data Mining and the Automatic Extraction of Scene Semantics in Ideological and Political Education

1.数据挖掘与思想政治教育场景语义的自动提取

Large models are essentially instance-based models, that is, "models composed of a large number of instances (language units and their degrees of association)." The instance-based nature of large models provides technical support for data mining of the scene of ideological and political education and the automatic extraction of semantics based on this. Among them, data mining is concentratedly manifested in aspects such as the formation of research questions, data selection, preprocessing, analysis, and communication presentation. At the level of research question formation, large models, relying on complex statistical induction and computational deduction, can not only achieve intelligent classification and clustering of theoretical issues in ideological and political education but also realize deep mining of meta-issues and specific issues in ideological and political education, pointing the way for the design and application of the educational scene. At the level of data selection, the scene of ideological and political education empowered by large models can conduct data mining based on educational requirements and problem research, systematically collecting text data, behavioral data, image data, etc., related to educational content, educational objects, and educational interaction according to the "instances" provided by the large model. At the level of data preprocessing, with the help of large models, input, cleaning, filtering, format reorganization, and content reference extraction of raw educational data can be achieved. Among them, data filtering and content reference extraction are particularly important, because data filtering can enhance the precision of educational scene analysis, and content reference extraction (such as symbolic representation of people, objects, and events) can simplify the complexity of educational scene analysis. At the level of data analysis, the scene of ideological and political education empowered by large models achieves diversified, multi-level, and multi-dimensional correlation analysis, such as word, grammar, and semantic analysis, spatial analysis, sequence analysis, similarity analysis, anomaly detection analysis, etc. These have positive significance for enhancing the foresight of the educational scene process and the meaning extraction of scene semantics. At the level of communication presentation, the results of data mining include the effects of scene education, potential impacts, and generated value. The empowerment of the scene of ideological and political education by large models not only makes personalized, real-time, and visualized analysis of data mining results a reality but also fully reflects the feedback benefits and practical value of data mining results.

大模型本质上是一种实例性模型，即“由大量实例（语元和语元关联度）构成的模型”。大模型的实例性为思想政治教育场景的数据挖掘以及在此基础上的语义自动提取提供了技术支持。其中，数据挖掘集中表现在研究问题的形成、数据的选择、预处理、分析以及交流呈现等层面。在研究问题的形成层面，大模型依托复杂的统计归纳与计算演绎，不仅能够实现思想政治教育理论问题的智能分类与聚类，而且能够实现对思想政治教育元问题与具体问题的深度挖掘，它为思想政治教育的场景设计与应用指明了方向。在数据选择层面，大模型赋能下的思想政治教育场景能够依托教育要求与问题研究进行数据挖掘，依据大模型提供的“实例”系统采集与教育内容、教育对象、教育交互等相关的文本数据、行为数据、图像数据等。在数据预处理层面，借助大模型可以实现原始教育数据的输入、清理、筛选、格式重编以及内容指代提取。其中，数据筛选和内容指代提取尤为重要，因为数据筛选可以提升教育场景分析的精准性，内容指代（如人、物、事件的符号表示）提取可以简化教育场景分析的复杂性。在数据分析层面，大模型赋能下的思想政治教育场景实现了多元化、多层次、多维度的相关关系分析，如语词、语法与语义分析、空间分析、序列分析、相似性分析、异常检测分析等，它们对提升教育场景过程展开的预见性与场景语义的意义提取有着积极的意义。在交流呈现层面，数据挖掘结果涵括场景育人的效果、潜在的影响、产生的价值等。思想政治教育场景的大模型赋能不仅使数据挖掘结果的个性化、实时化与可视化分析成为现实，而且充分体现了数据挖掘结果的反馈效益与实践价值。

2. Network Analysis and the Intelligent Evaluation of Scene Value in Ideological and Political Education

2.网络分析与思想政治教育场景价值的智能评估

The ideological and political education scenario manifests as a complex network of interpersonal interactions in social interaction, social dissemination, and social mobilization. The network analysis of large models provides a scientific method for deeply analyzing the complex operational state of ideological and political education scenarios. The network analysis of large models is centrally embodied in structural analysis, systemic characterization, validity analysis, and behavioral prediction. The first is structural analysis based on value assessment. The analysis of ideological and political education scenarios empowered by large models consistently follows the structural logic of “goal—choice—result—probability—utility,” aiming to assess the deep value of ideological and political education scenarios. Specifically, in the goal dimension, large-model analysis must focus on examining the adaptability of the education scenario to the external environment, the gap between the needs of the education target and social requirements, and the tension between the setting of the scenario context and educational demands. In the choice dimension, large-model analysis must not only consider diverse alternative plans but also achieve precise identification, personalized matching, and innovative optimization of these plans. In the result dimension, large-model analysis must not only fully consider the feedback nature of the operational results of the ideological and political education scenario but also integrate the feedback of results with the elements, processes, and even the entire scenario system for comprehensive investigation. The second is systemic characterization based on holistic emergence. The network analysis of ideological and political education scenarios empowered by large models possesses a holistic characteristic, meaning it does not assess the specific benefits of the ideological and political education scenario from the perspective of scenario elements but rather evaluates the emergent benefits of the scenario from the perspective of the interaction of scenario elements or the overall operation of the system. Through data annotation of education scenario elements, computational modeling of education scenario operation, and precise prediction of education scenario models, large models systematically characterize the holistic emergence of ideological and political education scenarios. The third is validity analysis based on remote interaction. Large models, with their memory advantages and computational functions, conduct validity analysis of remote interaction across time and space. In the temporal dimension, large models achieve cross-time dialogue for scenario remote interaction through their memory advantages for historical and expected data; in the spatial dimension, large models expand the scope of remote interaction analysis for education scenarios, enabling analyses such as the integration of virtual and real in remote interaction and the combination of explicit and implicit analysis. The fourth is behavioral prediction based on network correlation. Large models not only rely on data correlation to achieve stable analysis of scenario subject experience, connection quality, and emotional output status but also, based on the predictive function built on data correlation analysis, achieve potential analysis and sustainability analysis of scenario subject behavior.

思想政治教育场景在社会互动、社会传播与社会动员中展现为人与人之间的复杂性交往网络。大模型的网络分析为深入剖析思想政治教育场景的复杂性运行状况提供了科学方法。大模型的网络分析集中体现为结构分析、系统刻画、效度分析以及行为预测等方面。一是基于价值评估的结构分析。大模型赋能下的思想政治教育场景分析始终遵循“目标—选择—结果—概率—效用”的结构逻辑，旨在评估思想政治教育场景的深层价值。具体来讲，在目标维度，大模型分析要注重考察教育场景对外在环境的适应性、教育对象的需要与社会要求之间的差值以及场景情境的设置与教育需求的张力等。在选择维度，大模型分析不仅要考虑多样化的备选方案，还要实现对方案的精准识别、个性化匹配与创新优化。在结果维度，大模型分析不仅要充分考量思想政治教育场景运行结果的反馈性，而且要将结果的反馈与教育场景要素、过程甚至整个场景系统联系起来进行综合考察。二是基于整体涌现的系统刻画。大模型赋能下思想政治教育场景的网络分析有着整体性特征，即不是从场景要素出发评估思想政治教育场景的具体效益，而是从场景要素的相互作用或系统的整体运行出发评估思想政治教育场景的涌现效益。大模型通过对教育场景要素的数据标注、教育场景运行的计算建模以及教育场景模型的精准预测，系统刻画了思想政治教育场景的整体涌现。三是基于远程交互的效度分析。大模型以其记忆优势和计算功能在时空层面开展了远程交互的效度分析。在时间维度，大模型通过对历史数据与预期数据的记忆优势实现了场景远程交互的跨时间对话；在空间维度，大模型扩展了教育场景的远程交互分析范围，实现了远程交互的虚实共生分析与显隐结合分析等。四是基于网络关联的行为预测。大模型不仅依托数据关联性实现了对场景主体体验、连接质量以及情感输出状况的稳定性分析，而且依托数据相关关系分析基础上的预测功能实现了对场景主体行为的潜力分析以及可持续性分析。

3. Modeling and Simulation and the Intelligent Advancement of Ideological and Political Education Scenario Operation

3.建模仿真与思想政治教育场景运行的智能推进

The integrated intelligence and simulation intelligence of large models provide an intelligent foundation for the modeling and simulation of ideological and political education scenarios, which helps to deeply advance the complexity research of ideological and political education scenarios. The first is scientific modeling based on the integrated intelligence of large models. The modeling of ideological and political education scenarios is essentially a process of constructing a scenario computational model based on elements such as subjects, environment, and interaction rules. Through data aggregation and computational analysis of elements of ideological and political education scenarios, the integrated intelligence of large models achieves a systematic grasp of scenario modeling elements, enhances the scientific nature of rule-based modeling, interaction-behavior-based modeling, and behavioral-dynamics-based modeling, and provides a technical foundation for comprehensively grasping the educational mechanism of scenarios. In the subject dimension, large models not only rely on their data generation advantages to achieve precise grasp of the knowledge, emotions, intentions, behaviors, and future trends of education targets but also activate the autonomy of education targets as personalized collections. In the environment dimension, large models not only rely on digital interaction technology to achieve digital reproduction of real ideological and political education scenarios but also, relying on digital twin and native technologies, achieve the intelligent construction of environmental twins highly consistent with real ideological and political education scenarios. In the interaction rule dimension, the adaptive system of large models achieves the dynamic generation of scenario interaction rules, which drives the interaction mode of ideological and political education scenarios into a fluid multimodal construction. The second is simulation verification based on the simulation intelligence of large models. Ideological and political education simulation models do not always adapt to the needs of real education scenarios. Strengthening the feasibility analysis of ideological and political education scenario simulation has become a top priority for deepening the educational effect of scenarios. Large models, relying on their functional simulation advantages in code rehearsal, performance analysis, and parameter scanning, enhance the feasibility of education simulation models. At the code rehearsal level, large models can not only rely on their intelligent retrieval functions to achieve full-coverage reading and modular generation of code but also perform precise evaluation and intelligent reconstruction of code according to educational needs, improving the efficiency of scenario simulation testing. At the performance analysis level, large models, with their advantages in intelligent generation and analysis, achieve both quantitative, automated content analysis and information extraction, as well as intelligent comprehensive integration of performance analysis, greatly improving the quality of scenario simulation testing. At the parameter scanning level, parameters of education scenarios empowered by large models achieve intelligent generation and analysis, enhancing the testing of general attributes and the revelation of special attributes of ideological and political education scenarios. Among these, general attributes manifest as the stability and persistence of ideological and political education scenarios; special attributes manifest as the singularity effects and emergence effects of ideological and political education scenarios.

大模型的集成智能与模拟智能为思想政治教育场景的建模仿真提供了智能基础，它有助于深入推进思想政治教育场景的复杂性研究。一是基于大模型集成智能的科学建模。思想政治教育场景的建模本质上是基于主体、环境、交互规则等要素建构场景计算模型的过程。大模型的集成智能通过对思想政治教育场景元素的数据集合与计算分析，实现了对场景建模要素的系统把握，提升了基于规则建模、基于交互行为建模以及基于行为动力学建模的科学性，为全面把握场景育人机理提供了技术基础。在主体维度，大模型不仅依托其数据生成优势实现了对教育对象知情意行及其未来走向的精准把握，而且激活了作为个性化集合的教育对象的自主性。在环境维度，大模型不仅依托数字交互技术实现了现实思想政治教育场景的数字摹写，而且依托数字孪生与原生技术实现了与现实思想政治教育场景高度契合的环境孪生体的智能建构。在交互规则维度，大模型的适应性系统实现了场景交互规则的动态生成，它促使思想政治教育场景交互模式处于流动的多模态建构之中。二是基于大模型模拟智能的仿真验证。思想政治教育仿真模型并不总是适应现实教育场景的需求，加强对思想政治教育场景仿真的可行性分析，成为深化场景育人效果的重中之重。大模型依托其在代码演练、性能分析以及参数扫描层面的功能模拟优势，提升了教育仿真模型的可行性。在代码演练层面，大模型不仅能够依托其智能检索功能实现对代码的全覆盖读取与模块化生成，而且能够根据教育需要对代码进行精准评价与智能重构，提升了场景仿真检验的效率。在性能分析层面，大模型以其智能生成与分析优势，既实现了定量的、自动化的内容分析与信息提取，又实现了性能分析的智能综合集成，极大地提升了场景仿真检验的质量。在参数扫描层面，大模型赋能下的教育场景参数实现了智能生成与分析，增强了对思想政治教育场景一般属性的检验与特殊属性的揭示。其中，一般属性表现为思想政治教育场景的稳定性、持续性等；特殊属性表现为思想政治教育场景的奇点效应、涌现效应等。

4. Computational Experiments and the Virtual-Real Complementarity of Ideological and Political Education Scenario Improvement

4.计算实验与思想政治教育场景改进的虚实互补

Computational experiments empowered by large models aim to place educational simulation scenarios and educational real scenarios in a parallel perspective, achieving mutual complementarity of data, mining of laws, and predictive characterization of behavior in the bidirectional empowerment of virtual and real education scenarios. The first is data complementarity in bidirectional empowerment. Educational scenario experiments based on large models are conducted within a “parallel system.” Real scenario element data, scenario operation data, and scenario effect data are fed back in real time to the educational simulation scenario via large models in the form of “instances.” Computational analysis data from the educational simulation scenario are transmitted in real time to the real scenario via large models. The data of the two learn from and complement each other. This not only provides empirical evidence for finding the gap between virtual and real education scenarios and discovering operational problems in real ideological and political education scenarios but also offers practical references for scientifically evaluating the operational effects of ideological and political education scenarios and systematically improving their operational status. The second is law mining in bidirectional empowerment. With large models as the medium, educational scenario simulation and real education scenarios achieve scientific mining of scenario operational laws through dynamic comparison. On the one hand, ideological and political education scenario computation empowered by large models not only achieves data analysis of subject interaction behaviors and dynamic network collaboration status in education scenarios but also achieves data synthesis of scenario network structures and event evolution states, providing empirical support for exploring the operational laws of real education scenarios. On the other hand, through data generation, dynamic analysis, and activity summarization of real ideological and political education scenarios, large models can present the operational problems and contradictions of real education scenarios in a real-time, visualized, and personalized manner, providing a scientific orientation for the computational analysis of educational simulation scenarios and a practical basis for deeply mining the operational laws of education scenarios. The third is behavioral characterization in bidirectional empowerment. The behavioral characterization of real ideological and political education scenarios based on large models is a digital reproduction and replication of scenario subject behavior. It truthfully reflects the thoughts and behaviors of scenario subjects and the interaction status among subjects, providing a continuous stream of real material for the simulation computation of education scenarios. The simulation of education scenarios mediated by large models is not a direct copy of real scenario behavior but a presentation of computational models based on holism rather than reductionism, a characterization of scenario behavior oriented toward possibilities rather than certainties. It provides a computational basis for comprehensively grasping the complex laws of ideological and political education scenario operation and predicting the future development of ideological and political education scenarios.

大模型赋能下的计算实验旨在将教育仿真场景与教育现实场景置于平行视域之中，在教育场景的虚实双向赋能中实现数据的相互补充、规律的挖掘以及行为的预测刻画。一是双向赋能中的数据互补。基于大模型的教育场景实验是在“平行系统”中展开的，现实的场景要素数据、场景运行数据以及场景效果数据依托大模型会以“实例”的方式实时反馈给教育仿真场景，教育仿真场景中的计算分析数据会通过大模型实时地传输到现实的场景，两者的数据相互借鉴、相互补充，不仅为寻找虚实教育场景间的差距、发现现实思想政治教育场景的运行问题提供了实证依据，而且为科学评价思想政治教育场景的运行效果以及系统改进思想政治教育场景的运行状况提供了现实参考。二是双向赋能中的规律挖掘。以大模型为介导的教育场景仿真与现实教育场景在动态的比照中实现了场景运行规律的科学挖掘。一方面，大模型赋能下的思想政治教育场景计算不仅实现了对教育场景主体交互行为、动态网络协作状况等的数据分析，而且实现了对思想政治教育场景网络结构、教育场景事件演化状态等的数据综合，它为探索现实教育场景运行规律提供了实证支持。另一方面，大模型通过对现实思想政治教育场景的数据生成、动态分析与活动总结，能够实时化、可视化、个性化呈现现实教育场景的运行问题及矛盾，为教育仿真场景的计算分析提供了科学导向，为深入挖掘教育场景的运行规律提供了现实依据。三是双向赋能中的行为刻画。基于大模型的现实思想政治教育场景行为刻画是对场景主体行为的数字摹写与复制，它真实地反映了场景主体的思想和行为以及主体间的交互状况，为教育场景的仿真计算提供了源源不断的现实素材。大模型介导下的教育场景仿真并不是对现实场景行为的照搬复制，而是基于整体论而非还原论的计算模型呈现，是面向可能性而非确定性的场景行为刻画。它为全面把握思想政治教育场景运行的复杂规律、预测思想政治教育场景的未来发展提供了计算基础。

III. Innovation in Methods for the Evolution of Ideological and Political Education Scenarios in the Era of Large Models

三、大模型时代思想政治教育场景进化方法创新

Large models activate the systemic evolutionary attributes of ideological and political education scenes, such as self-adaptability, self-organization, self-replication, and self-creativity, thereby promoting the autonomous evolution, orderly evolution, iterative progression, and holistic emergence of ideological and political education scenes.

大模型激活了思想政治教育场景的自适应性、自组织性、自复制性、自创造性等系统进化属性，推动了思想政治教育场景的自主进化、有序进化、迭代演进以及整体涌现。

1. Activation of Self-Adaptability in Ideological and Political Education Scenes and Autonomous Evolution

1.思想政治教育场景的自适应性激活与自主进化

The self-adaptability of ideological and political education scenes is essentially the property of the educational scene autonomously adapting to the environment. It is manifested through the conscious interaction between the educational scene and the external environment, as well as the organic interaction among the internal elements of the educational scene. Large models, with their intelligent capabilities in connection, generation, and analysis, activate the self-adaptability of ideological and political education scenes, realizing their autonomous evolution. At the level of intelligent connection, the vast data parameters and natural language interaction advantages of large models endow them with a unique "search and aggregation" function, providing technical support for the timely emergence of ideological and political education scenes and the establishment of educational relationships on this basis. On the one hand, through precise and personalized searches of educational subject information, large models can enrich and optimize the content of educational scenes, activating the organic interaction between the educational scene and the educational subject. On the other hand, by aggregating viewpoints related to various questions and answers from educational objects, large models can provide scene information tailored to the internal needs of educational objects. At the level of intelligent generation, the advantages of large-scale corpora and correlation prediction in large models give them a powerful "content generation" function. This not only provides a continuous source of material for enhancing the understanding and interpretation capabilities of ideological and political education scenes but also enables the adaptive generation and precise推送 of educational scenes based on predictive analysis of the psychological needs of educational objects. Among these, adaptive generation manifests as interesting and enhanced scene design, as well as personalized scene generation that incorporates social and educational requirements; precise推送 manifests as accurate alignment with the needs of educational objects and precise presentation of educational scene information. At the level of intelligent analysis, the advantages of data mining, computational analysis, and evidence-based experimentation in large models endow them with a highly efficient "design optimization" function. This can perform structural analysis of educational scenes by absorbing and computing complex problems during the pre-training process, systematically integrate the value orientation of educational objects with scene supply and demand, and formulate scene strategies tailored to the needs of educational objects based on data abstraction, computational modeling, and experimental analysis of educational scenes. Furthermore, large models, relying on full-element and full-process evaluation based on correlation analysis of big data, provide an important basis for the selection and optimization of strategies for ideological and political education scenes.

思想政治教育场景的自适应性本质上是教育场景对环境的自主适应特性，它通过教育场景与外部环境间的自觉交互以及教育场景内部要素间的有机互动表现出来。大模型以其智能化的连接、生成以及分析属性激活了思想政治教育场景的自适应性，实现了思想政治教育场景的自主进化。在智能连接层面，大模型庞大的数据参数与自然语言交互优势使其具有独特的“搜索聚合”功能，它为思想政治教育场景的及时出场以及在此基础上的教育关系的建立提供了技术支持。一方面，大模型通过对教育主体信息的精准搜索与个性化搜索，能够丰富与优化教育场景内容，激活教育场景与教育主体的有机交互；另一方面，大模型通过对关涉教育对象各类提问与回答的观点聚合，能够提供适配教育对象内需的各类场景信息。在智能生成层面，大模型的大规模语料库优势与关联度预测优势使其具有强大的“内容生成”功能，它不仅为增强思想政治教育场景的理解与诠释能力提供了源源不断的素材基础，而且能够在预测分析教育对象心理需要的基础上，实现教育场景的适配性生成与精准性推送。其中，适配性生成表现为趣味化、增强性的场景设计与蕴含社会要求和教育要求的个性化场景生成；精准性推送表现为对教育对象需要的精准对接与教育场景信息的精准展现等。在智能分析层面，大模型的数据挖掘、计算分析以及循证实验等优势使其具有高效益的“设计优化”功能，它既能够以其预训练过程中对复杂性问题的吸纳与运算对教育场景作出结构性分析，又能够对教育对象的价值导向与场景供需作出系统性整合，还能够基于教育场景的数据抽象、计算建模、实验分析等制定适配教育对象需要的场景策略。此外，大模型依托大数据相关关系分析基础上的全要素与全过程评价，为思想政治教育场景策略的选择与优化提供了重要依据。

2. Activation of Self-Organization in Ideological and Political Education Scenes and Orderly Evolution

2.思想政治教育场景的自组织性激活与有序进化

The so-called self-organization of ideological and political education scenes refers to the system property whereby, "when certain parameter conditions are all met," the ideological and political education scene can operate spontaneously without external environmental influence, thereby promoting its own transition from disorder to order. Large models promote the orderly evolution of ideological and political education scenes by activating the "parameter conditions" of the scene, such as the interaction of scene elements and the integrated operation of the scene system. First, the activation of nonlinear interactions among educational scene elements. The unique dialogue, expansion, and learning attributes of large models activate the correlations among educational scene elements, providing endogenous momentum for the self-organized operation of the educational scene. In the dimension of dialogue attributes, large models, leveraging natural language processing technology and multilingual dialogue technology, strengthen the analysis of educational scene elements and the storage and updating of dialogue records, stimulating mutual recognition and response among educational scene elements. In the dimension of expansion attributes, large models can adjust computing resources based on changes in educational scene elements through powerful computer clusters, using their model reasoning capabilities to improve the precision and generalization ability of interactions among educational scene elements. In the dimension of learning attributes, large models can not only summarize the patterns and modes of nonlinear interactions among educational scene elements through data analysis but also achieve the transfer and application of these patterns. Second, the differential integration of the operation of the educational scene system. The operation of ideological and political education scenes is not always orderly. Large models, through the continuous construction of "instances" and the sublation and optimization of "events," enable the ideological and political education scene system to operate in an open and self-sufficient manner. On the one hand, large models can integrate "language units" related to information in ideological and political education scenes and, through correlation analysis among different "language units," construct "instances" that meet the needs of the scene system's operation, providing an open resource library for the self-organized operation of the ideological and political education scene system. On the other hand, the ideological and political education scene system empowered by large models always operates and unfolds following an "event-oriented" logic oriented toward possibilities, activating its own self-organization in the generation and sublation of "events" related to society, technology, and culture. Furthermore, by absorbing and integrating "events," large models inject vitality into the interactive role between the educational scene and various "events" related to education itself.

所谓思想政治教育场景的自组织性，指的是“当一定的参量条件都得到满足时”，思想政治教育场景在不受外部环境影响下能够自发地运行起来，进而促使自身从无序到有序的系统属性。大模型通过对场景要素相互作用与场景系统整合运行等场景的“参量条件”的激活，推动了思想政治教育场景的有序进化。一是教育场景要素的非线性作用激活。大模型独有的对话属性、扩展属性以及学习属性激活了教育场景要素间的相关关系，它为教育场景的自组织运行提供了内生动力。在对话属性维度，大模型借助于自然语言处理技术以及多语言对话技术，强化了对教育场景要素的解析以及对话记录的存储与更新，激发了教育场景要素间的相互识别与呼应。在扩展属性维度，大模型通过强大的计算机集群可以根据教育场景要素的变化状况调整计算资源，以其模型推理能力提升教育场景要素间相互作用的精度与泛化能力。在学习属性维度，大模型不仅可以通过对教育场景要素数据的分析，总结归纳教育场景要素非线性作用的规律与模式，而且可以实现规律的迁移应用。二是教育场景系统运行的差异性整合。思想政治教育场景的运行并不总是有序的，大模型通过“实例”的持续建构与“事件”的扬弃优化，使思想政治教育场景系统得以开放自足运行。一方面，大模型能够整合与思想政治教育场景信息相关的“语元”，并通过不同“语元”间的相关关系分析建构契合场景系统运行需要的“实例”，它为思想政治教育场景系统的自组织运行提供了开放的资源库。另一方面，大模型赋能的思想政治教育场景系统始终遵循面向可能性的“事件导向”逻辑运行展开，并在关涉社会、科技、文化等“事件”的生成与扬弃中激活自身的自组织性。进而言之，大模型通过对“事件”的吸纳整合为教育场景与关涉教育本身的诸类“事件”的交互作用注入了生机活力。

3. Activation of Self-Replication in Ideological and Political Education Scenes and Iterative Progression

3.思想政治教育场景的自复制性激活与迭代演进

From the information dimension, the self-replication of ideological and political education scenes refers to the property whereby educators use "genetic information" derived from the original scene to perform a replication function, thereby constructing new scenes. Large models activate the self-replication of ideological and political education scenes through the interpretation and sharing of "genetic information." First, the interpretation of "genetic information" and the self-replication of ideological and political education scenes. Strictly speaking, "genetic information" exists only within communication activities, that is, "only within the communication system composed of the three lower-level systems: the source, the channel, and the sink." Large models, with their precise grasp of the source, real-time control of the channel, and effective evaluation of the sink, can dynamically and systematically interpret the content of "genetic information." Moreover, they can place "genetic information" within an open interpretive structure of synchronicity and diachronicity for a "fusion of horizons" examination. Among these, synchronic interpretation places "genetic information" in the comparative context of old and new educational scenes, relying on large model analysis based on "temporal distance" to "filter" prejudices. Diachronic interpretation attempts to provide a possibility-oriented explanation of "genetic information" through dialogue and understanding between the present and the past, creating a precondition for the continued evolution of ideological and political education scenes. This is because a mechanistic interpretation of "genetic information" would lead to homogeneous scene replication, thereby causing stagnation in scene evolution. Second, the sharing of "genetic information" and the self-replication of ideological and political education scenes. Large models, with their characteristics of cooperative participation and integrated optimization, activate the sharing advantages of "genetic information," promoting the collaborative production and system upgrade of ideological and political education scenes. On the one hand, by activating the sharing nature and "mutual information" characteristics of "genetic information" that align with the learning needs of educational objects, large models not only encourage educational objects to autonomously engage in the construction activities of ideological and political education scenes but also stimulate their cooperative intentions, collaborative awareness, and collective intelligence. On the other hand, through the recombination and selection of internal elements of "genetic information," large models promote the evolution and upgrade of ideological and political education scenes from lower-level systems to higher-level systems. Among these, ideological and political education scenes as higher-level systems are not always stable; they may degrade into new lower-level systems due to the recombination of "genetic information." In this cyclical manner, ideological and political education scenes, empowered by large models, truly achieve autonomous iterative progression.

从信息维度看，思想政治教育场景的自复制性指涉的是教育者借助脱胎于原有场景的“遗传信息”发挥复制功能以实现新场景建构的属性。大模型通过对“遗传信息”的诠释与共享，激活了思想政治教育场景的自复制性。一是“遗传信息”的诠释与思想政治教育场景的自复制。严格地讲，“遗传信息”仅仅存在于通信活动之中，即“仅存于由信源、信道和信宿这三个下层系统所组成的通信系统中”。大模型不仅以其对信源的精准把握、信道的实时控制以及信宿的有效评估，能够对“遗传信息”的内容作出动态的、系统的诠释，而且能够将“遗传信息”置于共时性与历时性的开放性诠释结构中进行“视界融合”考察。其中，共时性诠释将“遗传信息”置于新旧教育场景的比照语境之中，并依托“时间间距”基础上的大模型分析来“过滤”前见。历时性诠释试图在当下与过去的对话理解中对“遗传信息”进行面向可能性的解释，它为思想政治教育场景的继续进化创造了前提，因为关于“遗传信息”的机械化诠释会造成同质化的场景复制，进而引发场景进化的停滞。二是“遗传信息”的共享与思想政治教育场景的自复制。大模型以其合作参与和整合优化特性激活了“遗传信息”的共享优势，推动了思想政治教育场景的协作生产与系统升级。一方面，大模型通过对接教育对象学习需要的“遗传信息”的共享本性与“互信息”特性激活，不仅能够促使教育对象自主投入到思想政治教育场景的建构活动中来，而且激发了教育对象的合作意向、协作意识与集体智慧。另一方面，大模型通过对“遗传信息”内部元素的重组和选择，促使思想政治教育场景从低层次系统向更高层次系统演变升级。其中，作为高层次系统的思想政治教育场景并不总是稳定的，它会因“遗传信息”的重组而降阶为新的低层次系统。如此循环往复，思想政治教育场景在大模型的赋能下真正实现了自主迭代演进。

4. Activation of Self-Creativity in Ideological and Political Education Scenes and Holistic Emergence

4.思想政治教育场景的自创造性激活与整体涌现

The self-creativity of ideological and political education scenes is the attribute of self-generation from nothing that educational scenes manifest in the course of complex operation. Large models activate the self-creativity of ideological and political education scenes through open-source code, information integration, and environmental shaping, making the overall emergence of the ideological and political education scene system a reality. Among these, “the larger the model, the more layers, the greater the number of parameters, the more fully internal complexity and diversity develop, and the greater the opportunity for emergence.”[8]8 This quotation is from a Chinese source discussing the scaling laws of large models, emphasizing that increased model size, depth, and parameter count enhance internal complexity and the likelihood of emergent properties. First, open-source code and the self-creativity of ideological and political education scenes. Through open-source code and model reconstruction, large models not only meet the experiential demands of educational subjects and the developmental needs of ideological and political education scenes but also stimulate the collective creative will and value-sharing awareness of educational subjects, injecting vitality into the self-creativity of ideological and political education scenes. Specifically, open-source code lowers the application threshold of large models, endowing educational subjects with the ability to autonomously create scenes, and, through their generative synergy in text, images, video, games, and even code, promotes the autonomous production of ideological and political education scenes. Second, information integration and the self-creativity of ideological and political education scenes. The interactions among the elements of ideological and political education scenes, as well as the interaction between the scene system and the environment, are all represented and identified through information. As a systematic information integration platform, large models activate the creativity and integrative power of information. They not only activate the creative capacity of information through the organic operation of ideological and political education scenes in dimensions such as information collection, transmission, communication, processing, storage, utilization, and translation, but also, through the social integration of interaction rules, value concepts, cultural rituals, etc., within educational scenes, dissolve various boundary forces of educational scenes and activate the integrative power of the scene system, thereby promoting the autonomous evolution of ideological and political education scenes. Third, environmental shaping and the self-creativity of ideological and political education scenes. The creation of ideological and political education scenes is always situated within the interaction between the scene system and the environment. Since the environment evaluates and selects the generation, operation, and evolution of the educational scene system, educational scenes must undergo self-evolution according to the standard of adapting to the environment. By shaping an environment full of diversity, difference, and complexity, large models not only stimulate the motivation for self-transformation and self-improvement of ideological and political education scenes but also, on this basis, promote the overall emergence of the ideological and political education scene system.

思想政治教育场景的自创造性是教育场景在复杂性运行中体现出来的从无到有的自我生成属性。大模型通过代码开源、信息整合与环境塑造激活了思想政治教育场景的自创造性，使思想政治教育场景系统的整体涌现成为现实。其中，“模型越大，层数越多，参数数量越多，内部的复杂性、多样性就越充分，涌现机会更大”。一是代码开源与思想政治教育场景的自创造。大模型通过代码开源及其模型重构，不仅契合了教育对象的体验诉求与思想政治教育场景的发展需求，而且激发了教育对象的集体创造意愿与价值共享意识，它们为思想政治教育场景的自创造注入了活力。具体来讲，代码开源降低了大模型的应用门槛，赋予了教育对象自主创造场景的能力，并以其在文本、图像、视频、游戏甚至代码方面的生成合力推动了思想政治教育场景的自主生产。二是信息整合与思想政治教育场景的自创造。思想政治教育场景要素间的相互作用以及场景系统与环境的交互作用都通过信息来表征和识别。大模型作为系统的信息整合平台激活了信息的创造力与整合力，它不仅通过思想政治教育场景在信息采集、传递、通信、加工、存储、利用、转译等维度的有机运作激活了信息的创造能力，而且通过教育场景中交往规则、价值理念、文化仪式等的社会整合，实现了对教育场景诸类边界力量的消融以及场景系统整合力量的激活，推动了思想政治教育场景的自主进化。三是环境塑造与思想政治教育场景的自创造。思想政治教育场景的创造始终置身于场景系统与环境的交互作用之中。由于环境对教育场景系统的生成、运行与演化具有评价与选择作用，教育场景必须以适应环境的标准进行自我进化。大模型通过对充满多样性、差异性与复杂性环境的塑造，不仅激发了思想政治教育场景自我改造与自我完善的动力，而且在此基础上推动了思想政治教育场景系统的整体涌现。

(The authors are, respectively, an associate professor at the School of Marxism, Jiangxi University of Finance and Economics, and a master’s student at the School of Marxism, Jiangxi University of Finance and Economics.)

（作者系江西财经大学马克思主义学院副教授；江西财经大学马克思主义学院硕士研究生）

Online editor: Ma Jingren

网络编辑：马京仁